Applied Physics Research; Vol. 16, No.

1; 2024
ISSN 1916-9639 E-ISSN 1916-9647
Published by Canadian Center of Science and Education

Dual-Sided Rotor Design for Performance Boost of Synchronous

Reluctance Motors in Electric Vehicles
Luis Teia1

1
von Karman Institute for Fluid Dynamics, Chaussée de Waterloo 72, 1640 Rhode-Saint-Genèse, Belgium
Correspondence: Dr Luis Teia, von Karman Institute for Fluid Dynamics, Chaussée de Waterloo 72, 1640 Rhode-Saint-
Genèse, Belgium. E-mail: [email protected]

Received: October 21, 2023 Accepted: December 6, 2023 Online Published: February 7, 2024
doi:10.5539/apr.v16n1p20 URL: https://doi.org/10.5539/apr.v16n1p20

Abstract
Increasing environmental awareness is pushing the design of electic motors to favor none rare-earth solutions (i.e., with-
out permanent magnets), and one such example is the SyncRM 2 (or concentrated-coiled SRM2) being proposed for the
electric hybrid automobile Toyota Yaris. Following on an already established line of research on this topic, this article
proposes a new design that re-assigns most of the magnetic material in the stator to the rotor — resulting in the Dual-sided
SyncRM (a variant of the SRM2). The detrimental effect (caused by the extra gap) of slightly reducing the aligned induc-
tance is overwhelmingly outweighed by the beneficial effect of drastically reducing the unaligned inductance. Extensive
back-to-back FEMM analysis was conducted, where the recomputed SRM2 matches previous research, providing con-
fidence to the favorable predictions of the Dual-sided SyncRM. Both performances are compared, with the venue being
available for download on an open-source database. A realistic photo-rendered three-dimensional model is displayed and
also available. An important outcome is the Dual-sided SyncRM torque (and power) increased by 29% (with respect to
the SRM2), achieving a saliency ratio of 10 and an efficiency boost to 91% (at the rated operational speed of 1200rpm).
Keywords: switched reluctance motor, magnetic FEM, analytic method, torque, power, efficiency
The hybrid vehicle Toyota Yaris (Figure 2a) runs on an electric motor possessing permanent magnets (Figure 2b) [Takeno
et al, 2012]. There is a contemporary global effort to reduce the usage of rare-earth materials that has fueled research to
find alternative solutions. A recent effort resulted in the formulation of an alternate design (Figure 2c) that uses solely the
reluctance torque effect to operate [the Synchronous Reluctance Motor 2, or SRM2 as defined by Takeno et al (2012)].
Upon numerical simulations and experimental tests, this was proven to be an effective solution. Then building on this
foundation, Stuikys and Sykulski (2020) created an effective and traceable low-order model (assisted by FEM magnetic
numerical modeling) that predicted the SRM2 performance with impressive accuracy.

Figure 1: (a) The hybrid automobile Toyota Yaris, (b) its electric motor and (c) a possible SyncRM alternative

Continuing on this approach, the present article builds on this collective research effort, and uses this low-order model
to modify the stator and rotor design towards boosting its torque, power and efficiency. Over the past 50 years, research
studies (Menzies 1972, Landislav et al 2020, Ionel and Popescu 2011) involving SR motors designed them such that the
stator’s magnetic mass was equivalent (or larger) than that of the rotor. This over unity stator-to-rotor mass ratio greatly
influences the magnetic circuit impedance response, as the rotor turns between the aligned to the unaligned cases, and
consequentially its performance.

20
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

1. Hypothesis
A larger co-energy production (and thus torque/power per motor step rotation) is potentially achievable by minimizing the
mass portion of magnetic conductor in the stator (that is, by re-assigning this mass to the rotor), such that the impedance
when the rotor is unaligned is dramatically reduced (with an acceptable collateral reduction of the aligned impedance).
2. Theory
The SRM2 is a radial flux synchronous reluctance motor possessing at the center an aluminium shaft, a steel rotor with
18 physical poles, and a surrounding steel stator with 12 physical poles (Figure 2a). The gap between the stator and rotor
is 0.5mm wide. Each stator pole has concentrated coils composed of twenty two sets of 17 parallel connected turns of
AWG wire with 0.6mm internal diameter. They are driven by a three-phased electric bus controlled via a Pulse-Width
Modulation (PWM) scheme (characteristic for this type of motor). This means that sets of 6 coils disposed in a hexagonal
manner are driven by one of the phases, creating six electromagnetic poles that (with alternation from one phase to the
next) creates the effect of a rotating magnetic field. The rotor cavities between the stator and rotor are filled with air, as
is the surroundings of the motor/stator (representing an important boundary condition to be implemented later during the
numerical analysis).
The proposed design modification (hereafter termed Dual-sided SyncRM shown in Figure 2b) circumferential splits the
stator at the radially outer extremity of the windings, and re-assigns that outer part of the magnetic circuit to the rotor
(which is hence forth termed the rotor outer ring). This has the consequence of generating an outer gap (in addition to the
inner gap), and of making the motor slightly bigger (i.e., the outer diameter increased). However, the modification of the
motor did not affect its inner dimensions, that is the diameter of the shaft remains the same, and so does the (now termed)
rotor inner ring (where before, it was just the rotor in Figure 2a), the size and locations of the windings is also the same,
and so is the location and size of the stator poles. It is worth mentioning that the modified design is just a first attempt
that is by no means optimized; there is plenty of room for improvement. It is assumed that the Dual-sided SyncRM will
operate via the same power bus and PWM-controller scheme as the SRM2 [for further details, please see Stuikys and
Sykulski (2020)].

Figure 2: Axial midplane cut-out of (a) SRM2 and (b) Dual-sided SyncRM

A 3D model of the Dual-sided SyncRM is shown in Figure 3, and available (FCStd1, Step, Stl) to download here at this
author’s open profile page. This was built with the open-source software FreeCAD, and professionally photo-rendered
using the open-source software CADrays. It’s key characteristics are now briefly explained. The stator is composed
of a rear structural disc-like feature that fixes onto the vehicle interface, and to which (on the other side) the poles are
connected (Figure 3a). In turn, these stator poles slide inside the Dual-sided rotor, in itself composed of an inner and outer
ring linked by structural connectors. The windings are virtually the same as the SRM2, with slight modifications being
made at each extremity (necessary to hold the windings in place). As shown in the zoom at the lower left corner of Figure
3a, the individual wires are turned around the open slot/gap at the front end and at the closed slot/gap at the back. These

21
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

prevent the wires from rising above the inner and outer surface of the stator pole (which would result in a clash with the
rotor), allowing the rotor to turn freely.

Figure 3: Dual-sided SyncRM: (a) assembly perspective view, (b) cut-out perspective view and (c) cut-out front view

22
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

A back bearing (Figure 3a left) connects the (fixed) rear disc-like feature of the stator to the shaft (that is, the outer surface
of the bearing is fixed from rotating, and the inner surface turns with the shaft) [this is more readily visible at the back
of the half model in Figure 3b]. A front bearing provides the second support for the shaft (and thus also to the rotor).
Similarly, the outer surface of the front bearing connects to a fixed structure of the vehicle. The rotor is geared to the
shaft (via three angularly equidistant slots on the rotor that slide into three corresponding angular protrusions on the shaft
— visible on Figure 3b). The rotor transmits its loads to the shaft, that in turn drives the wheel and/or axle (to which is
connected via the inner teeth) [as the half model front view in Figure 3c illustrates]. From a manufacturing perspective,
and while it may be complex to execute, it is entirely possible to machine the Dual-sided rotor as a single piece via a series
of combined axial translation and rotational cuts. Alternatively, the single piece could be manufactured first via casting,
which would be followed by precise machining of the critical surfaces (i.e., the cylindrical inner surface interface to the
shaft and the poles’ surfaces interfacing the inner and outer gaps) to the required precision. If extra stiffness (to the stator
poles) is required, the tip of the stator poles can be connected via a ring (not shown in Figure 3), where care must be taken
for this new ring not to interfere with the rotor. Making both the rotor and stator as single pieces makes it is easier to
control the tolerances required to achieve the tight gaps, both radially inner and outer. Achieving the necessary gap width
during assembly and operation is priority, and to that effect, the physical geometry of the support connectors in the rotor
can (if necessary) be made of another shape (e.g., with angled corners) if it proves to be more convenient for manufacture.
3. Analysis Method
The impact of the modified design on motor performance is computed using the approach provided by Stuikys and
Sykulski (2020), encompassing a new low-order model applicable to any generic radial-flux SyncRM. Several reasons
lead to this choice. One reason is that by using vector theory, the area of the quadrilateral between the aligned and
unaligned flux-linkage responses (to changing operating current) determines directly the conversion energy W ′ , which in
turn translates into reluctance torque and electromagnetic power (this is further explained later in the Results section).
Another reason is that this method is validated against experimental data provided by Takeno et al (2012), which yielded a
good correlation. Moreover, it is well documented and clearly traceable (making it transparent), allowing identification of
how it works and of its limitations. To this effect, the present research charted numerically the flux-linkage performance,
both for the original SRM2 (Figure 4a) and novel Dual-sided SyncRM (Figure 4b). The planar electromagnetic simulation
of the motor was conducted using the open-source software FEMM, with the motor cross section drawn using the open-
source software NanoCAD (Arslan, 2021). Each mesh comprise of ∼ 35500 cells for the SRM2 and ∼ 48400 cells for
the Dual-sided SyncRM. Simulations were conducted for various rotor angular positions, namely 0, 2, 4, 6, 8, 10, 12 and
15 degrees (between stator and rotor pole axis). The results are available as DWG and FEMM files for download at this
author’s profile page on the open-source platform Figshare, for both the SRM2 and the Dual-sided SyncRM.

Figure 4: FEMM mesh for the (a) SRM2 and (b) Dual-sided SyncRM

23
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

The motor depth was set (at the FEMM ”Problem definition” menu) identical to that of the SRM2 (or 135mm) [1,2],
allowing for a compliant comparison between the two machines. Material seeds were added to each corresponding region.
Practically all the materials necessary for the calculations were found in the FEMM software library, except for the wire
which had to be adapted via a simple modification to the existing AWG wires. The SRM2 had each stator pole coiled with
17 parallel-connected wires (each with a diameter of d=0.6mm) with 22 turns (around the pole) each, stacked on top of
each other. The approach
√ used to √ model this in FEMM was to lump the 22 turns into a single turn of an equivalently larger
diameter wire (D = N × d = 22 × 0.6 ≈ 3mm, having the same group crossectional area and thus transporting the same
current —- skinning effects are not considered here), and coil it 17 times around the pole. An electrical circuit named
”R” was created and assigned to the wiring region at either side of the top dead center physical stator pole. The left side
presents the positive (into the page) bundle of 17 modified AWG wires (connected in series), while the right side presents
the negative (out of the page) portion of those same 17 modified wire bundles. Since in this simulation only phase 1 is
active (refer back to Figure 2a), the remainder wires (of phase 2 and 3) were assigned the same wire material but without
any circuit. Unfortunately, at the time of creation of this research, data on the steel 10JNX900 (Misao et al, 2005) was
not known. Instead, steel M-15 was used (for both the stator and rotor), as it was readily available in the FEMM library
[for additional information on steel M-15, see for example AKSteel (2007)]. As a follow-on work, the high performance
steel reference 10JNX900 [Youyou Technology Co. Ltd, 1993] (used for the original SRM2 motor) could be programmed
into the FEMM files [most importantly the B-H curve — for more information see the example of Wakisaka et al (2013)]
and assigned as a new rotor-stator material for a rerun, and subsequent comparison to the results from this research. In
turn, the lightweight and structurally-strong non-magnetic material aluminium 1100 is used for the shaft. As a boundary
condition, a region of air was added to the outer diameter of the stator (providing the magnetic insulation of the stator
from the rest of the environment). Seeds of the material Air were also allocated to the cavities in between the rotor and
the stator, including their inner and outer gaps. Similarly to the SRM2, the width of both inner and outer air gaps were
set to 0.5 mm each. Additional key dimensions of the motors are as follows: Stator outer diameter is 234.5mm, while
the inner diameter is 178.9mm. The rotor outer diameter is 296mm and inner diameter 102.6mm. It is likely that the
magnetic flux-linkage (a parameter critical to determine the Co-Energy W’) is being computed (by FEMM) via following
the flux lines until they close a loop, following by integrating along that path [as explained in Guilera (2018)]. This is
done for each circuit (”R” being one already mentioned, and another called ”B”), resulting in the values exported from
the software. The total flux-linkage Ψ of the motor is the sum of those computed for each circuit (i.e., ”R” plus ”B”).
4. Results
The Magnetic Field Intensity B distribution within stator and rotor for both motors are shown as color plots in Figure
5. The predicted magnetic loops (in Figure 5) are clearly seen all around the rotor-stator assembly. High regions of
flux are observed around the coils, with the flux direction alternating as expected between each of the six rotor segments.
Comparing the aligned cases of the SRM2 (Figure 5a) and the Dual-sided SyncRM (Figure 5b) shows that the flux density
is maintained high at the stator pole and inner rotor pole in both cases, but drops visibly (radially immediately after the
stator pole) because of the introduction of the new (outer) gap. This drop is to be expect since air has a much lower
magnetic permeability than steel. Therefore, qualitatively the extra gap is expected to have a noticeable impact on the
induction for the aligned case. However, comparing the flux intensity for the unaligned cases between the SRM2 (Figure
5c) and the Dual-sided SyncRM (Figure 5d) shows that first is much higher. This occurs because for the Dual-sided
SyncRM the source of magnetism (i.e., the stator pole) is now much more isolated from the rest of the magnetic material,
resulting in a substantial reduction in induced magnetic flux on all of the steel material (both rotor and stator). That is, the
magnetization effect from the coils is much more isolated, and thus weak. Qualitatively, this isolation of the stator poles
is expected to have a substantial impact on the overall induction for the unaligned case. Quantitatively, the performance
of both the SRM2 and Dual-sided RM is calculated by applying the method from Stuikys and Sykulski (2020), which is
now explained. The total electromagnetic power P provided by torque T at a rated angular speed ω is given as
 q.Nr 
P=T ×ω with T = W′ × ×R (1)
2π
where Nr is the number of physical poles of the rotor, and q is the number of electric phases. The factor R removes
the overlap between subsequent rotational unaligned-aligned segments (where the electrical input switches sequentially
between the 3 phases), being defined as a factor dependent on the segment angle β s , the number of stator poles N s and the
number of rotor poles N s , resulting in

1  360 360  1   360 360 

R=1+ βs − − =1+ 10.5 − − = 1.0476 (2)
βs Nr Ns 10.5 12 18

24
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

Figure 5: Magnetic flux distribution for the [0 deg] (a) SRM2 and (b) D-SRM; and [15 deg] (c) SRM2 and (d) D-SRM

The magnetic co-energy W ′ is computed from the mapping of flux-linkage response (as the rotor transits from unaligned
to aligned to the stator) for varying electrical current. In fact, the co-energy is the area in between the response lines
from the completely unaligned to completely aligned cases. The previous qualitative magnetic flux density change with
stator-rotor alignment in Figure 5 (for both motors) are now seen quantitatively in Figure 6 (for a wider range of rotor
angles and currents). Figure 6a shows the flux-linkage Ψ response (a parameter directly proportional to the flux density
ϕ at the stator pole) for varying rotor mechanical angle for the SRM2, and Figure 6b for the Dual-sided SyncRM). Both
cases present a similar response, in that a maximum is observed when aligned (i.e., 0 deg) that gradually reduces towards
unalignment (i.e., towards 15 deg), where current plays in general an amplification effect of the response. As noted earlier,
the fact that the Dual-sided SyncRM breaks the magnetic circuit on either side of the windings (effectively isolating them
during unalignment), makes the drop in flux-linkage in the unaligned angles (in Figure 6b) much more pronounced than
with respect to SRM2 (in Figure 6a). Figure 6c shows the flux-linkage Ψ response for varying current for SRM2, and
Figure 6d for Dual-sided SyncRM. This is the same data as in Figures 6a and 6b, except they are plotted for varying
current — this will be more usefull when later we compute electromagnetic torque and power. In Figures 6c and 6d, both

25
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

motors present a linear response up to a current of about 80A, where the steepness of the response is very much dependent
on the stator-rotor alignment. However, most engines operate at much higher currents than 80A (the SRM2 operates at
a rated current of 320A)[Takeno et al 2012, Stuikys and Sykulski 2020]. Beyond 80A, the effect of magnetic saturation
starts to appear, resulting in a non-linear response (which is connected to the non-linear B-H curve inherent to the selected
magnetic material for the stator and rotor) — a non-linear effect that becomes more pronounced as the rotor aligns with
the stator (i.e., when the magnetic flux density intensifies). Note that the proposed design modification has substantially
reduced the inclination of the linear response for the unaligned case (i.e., 15 deg) from the SRM2 (lowest line in Figure
6c) to the Dual-sided SyncRM (lowest line in Figure 6d).

Figure 6: Flux-linkage map [versus mechanical angle] (a) SRM2 and (b) Dual-sided; and [versus current] (c) SRM2 and
(d) Dual-sided

Computation of the Co-Energy W ′ requires only accounting the response from the extreme cases [i.e., when the rotor
is aligned (0 deg) and unaligned (15 deg)], as previously shown in Figure 6c for SRM2 and Figure 6d for Dual-sided
SyncRM. From a practical perspective, subsequent analysis requires these sets of two lines to be replotted in Figure 7, one
set for SRM2 (in black) and another for Dual-sided SyncRM (in red). From this raw data (in Figure 7), additional pertinent
information on motor performance can be extracted. To make a distinction on what information belongs to which motor,
the parameters associated with the Dual-sided SyncRM have henceforth an apostrophe at the end (all others, including
those that are both common to the two motors and that pertain solely to SRM2, do not have this apostrophe). Disregarding
commutation for now (i.e., disregarding the effect of reducing current prior to stator-rotor pole alignment, thus preventing
negative torque), it can be seen that the design modification has substantially increased the conversion energy from the
SRM2 (i.e., the black area WS′ RM2 delimited by the quadrilateral OABD) into that of the Dual-sided SyncRM (i.e., the red

26
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

′
area WDual−S delimited by the quadrilateral OA′ B′ D′ ). The slight reduction in area caused by the (extra) radially outer
gap (i.e., the area ODBB′ D′ ) is superseded by the larger increase in area caused by the reduced unaligned induction of
the magnetic circuit (i.e., the area OA′ A). Prior to magnetic saturation in the stator and/or rotor, the flux-linkage response
to increasing current is practically linear (as shown by lines OD for SRM2 and OD′ for Dual-sided SyncRM). When
saturation occurs (at around 80A in SRM2), the slope reduces drastically presenting a non-linear response. This is to
be expected from the definition of flux-linkage Ψ = N.ϕ, where N is the number of coils around the stator and ϕ is
the magnetic flux density, reflecting the B-H curve behavior of the stator/rotor ferromagnetic material [being for steels
tipically non-linear (Wakisaka et al 2013, YouYou Technology Co. Ltd. 1993)]. Linear trendlines were ploted for each of
the three cases, both for the SRM2 (in black) and for the Dual-sided SyncRM (in red), including the respective equations
and R2 levels of confidence. The inclination of each trendline is — by definition — the inductance value L for each
′
stator-to-rotor position, namely for example the Dual-sided SyncRM: Luu for when the circuit is unsaturated and the rotor
′ ′
is unaligned, Lua when unsaturated and aligned, and finally L sa when saturated and aligned.

Figure 7: Co-energy assessment for both the SRM2 (in black) and Dual-sided SyncRM (in red)

However, commutation is typically employed to avoid unwanted production of negative torque. Thus, for the Dual-sided
SyncRM the rotation of the rotor through an angular sector is described (in Figure 6) by the yellow dashed lines with
vectorial path OA′C ′ E ′ (instead of the uncommuted path described by lines surrounding the red area OA′ B′ D′ ). That is,
for the Dual-sided SyncRM, the rotor path starts at point O in a completely unaligned state, being followed by a ramp-up
towards the rated current ir . Then this constant rated current is applied for a period of time, during which the unaligned
magnetic circuit drives the rotor to rotate further. As the rotor pole approaches alignment with the stator pole, the current
starts to reduce, and commutation occurs at point B’ (instead at point B, where it would be completely aligned). If the
rated current ir acted all the way until the rotor was fully aligned with the stator, then instead the Dual-sided SyncRM
would follow the path OA′ B′ D′ , and in the case of the SRM2 it would follow the path OABD. According to Eq.(1), in
order to proceed with the computation of the produced torque and EM power it is necessary to compute the Co-Energy
W ′ (i.e., the commuted area in the general quadrilateral OA′C ′ E ′ in Figure 7 for the Dual-sided SyncRM), which is
determined via the mathematical application of vector theory. A brief summary is now given. Vector theory (Shapiro and
de Berredo-Peixoto, 2013) states that the area of a general quadrilateral with vertices (x1 , y1 ) = (0, 0), (x2 , y2 ) = (ir , Ψuu ),
(x3 , y3 ) = (ir , Ψ sa ) and (x4 , y4 ) = (i s , Ψua ) [listed counterclockwise along its perimeter] is given by
( ) (  )
1 0 ir ir ir ir is is 0 1   
W′ = + + + = ir Ψua − Ψuu − i s − ir Ψ sa (3)
2 0 Ψuu Ψuu Ψ sa Ψ sa Ψua Ψua 0 2

27
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

where the matrices in between the modulus are determinants, solved to be equal to the subtraction of the cross-products.
Expanding this (by replacing the flux-linkage definitions Ψua , Ψuu and Ψ sa in Figure 7) gives
( )
1
W =′
(L sa − Luu )ir + (Lua − L sa )ir i s + cΨ s ir − cΨ s i s
2
(4)
2

Careful observation of Figure 7 shows that the saturation current i s is deduced as

Ψs 0.35575
is = = = 79.6A (5)
Lua − L sa 0.0049 − 0.00043

According to Stuikys and Sykulski (2020), Eq. (4) can be further expanded to become a function of angles, voltages,
currents and inductances, as

(Vrms(PW M) c.βs
ω + (Luu − L sa ).ir )
2)
(
1 c.β s
W = ′
2.Vrms(PW M) .ir + (Luu − L sa )ir −
2
(6)
2 ω Luu − L sa

For a designer of electric motors, Eq.(6) can be used for example to perform sensitivity studies, or to taylor the machine
to a specific torque and/or power requirement. For those who wish to understand and improve this method, here is a brief
explanation on how this method was derived. For additional details, please refer to their original publication. The present
author started by applying Eq.(6) to the reconstructed 2D CAD model of the original SRM2, which gave a very close
approximation of the results given by Stuikys and Sykulski (2020) [summarized later in Table 1, which can be compared
by referring to their original article]. Upon re-application and verification of this method to the baseline SRM2, it was
subsequentially applied to the Dual-sided SyncRM, with the key calculations for the later being presented below. The
target rotational speed of the original SRM2 is 1200 rpm, which for a machine with Nr = 12 physical rotor poles gives

rpm × number of poles 1200 × 12

f = = = 60Hz (7)
120 120
The argument could be done inversely, in that for a target operating mains frequency of 60Hz, and a SR motor with 12
physical rotor poles, the operational rpm will be 1200. In radians per second, this translates to a rotational speed ω of

1200
ω= × 2π = 125.7rad/s (8)
60
The computation of the commutation angle takes advantage of the fact that the unit of flux-linkage, the Weber, is also
expressible as Volts-second, which provides (in conjunction with the angular speed ω) a direct bridge between the angular
position of the rotor θ and the magnitude of the flux-linkage Ψ. Thus, Figure 9 suggests that the commutation angle θCE
is equivalent to the time taken for the current to fall from point C’ to E’ (a vertical flux-linkage drop equal — but offset —
to that from point B to D) [this may be difficult to vizualize now, but it will become cleared later when analyzing Figure
9] . This flux-linkage vertical drop occurs at an angular rotation ω and bus voltage VDC , resulting in

Ψ sa (@B) − Ψ sa (@D)
" # " #
0.49335 − 0.389972
θCE = ×ω= × 125.7 = 1.49deg (9)
VDC 500

Another important parameter to compute is the commutation factor c, which is a multiplier coefficient that accounts for
the aforementioned period of ramping down of the current (from point C’ to E’) in the OA′C ′ E ′ cycle, and is given as

θCE 1.49
c=1− =1− = 0.86 (10)
βs 10.5

The rms voltage Vrms(PW M) (across the windings) necessary to counteract the back-electromotive force (at the rated speed)
is found in the definition of the saturation flux-linkage constant Ψ s (i.e., where the line D′ B′ crosses the vertical axis in
Figure 7), resulting in

βs ir
Ψ s = Vrms(PW M) − (L sa − Luu ) (11)
ω c

28
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

which when re-arranged (and substituting the values in Figure 7), becomes

ir ω
! !
320 125.7
Vrms(PW M) = Ψ s + (L sa − Luu ) = 0.35575 + (0.0005 − 0.00043) = 226V (12)
c βs 0.86 10.5

Finally, all the information necessary to compute the co-energy W ′ is now available. Applying the above values to Eq.(6)
gives

 2
1
"
(0.86)(10.5)   226 (0.86)(10.5)
125.7 + (0.0005 − 0.00043).320 #
W ′ = 2.(226) .(320) + 0.0005 − 0.00043 (320)2 − ≈ 84J
2 125.7 0.0005 − 0.00043

Therefore, the electromagnetic torque and power [computed from Eq.(1)] becomes

(6)(12)
T = 84 × 1.0476 = 502.3Nm giving P = (502.3)(125.7) = 63126W
2π
Computations of the above equations were also done for the case of the SRM2, and took into account the equivalent
parameters and variables as defined previously in Table 2 [e.g., in line 21, the delta flux-linkage Ψ(@B′ ) − Ψ(@E ′ ) =
0.103378 for the Dual-sided SyncRM is in fact Ψ(@B) − Ψ(@D) = 0.081236 for the SRM2 (in accordance to Figure
7)]. Note that some parameters are common to both motors. Figure 8 shows the computed speed-torque and speed-
power characteristic curves [as per Qi et al (2019)] for both SRM2 and Dual-sided SyncRM. The later motor (Dual-sided
SyncRM) shows an improvement performance, in that it can provide the original target SRM2 torque of ∼ 400N.m up to
1500rpm (higher than the original rated 1200rpm).
Efficiency is given as the ratio of produced electromagnetic power Pem to consumed average electric power Pel . The EM
power has already been computed, and the average electric power Pel = Vav .Iav is determined by analyzing the current
(solid red lines in Figure 9) and voltage (solid blue lines in Figure 9) response as the rotor turns from the unaligned to
aligned case. This then yields the average current Iav (dashed red line) and average voltage Vav (dashed blue line) for
a complete rotational segment (from unalignment to alignment of rotor-to-stator poles). The flux-linkage characteristics
can be converted into time periods of seconds (for each of the sectors in Figure 9) via the definition of Weber = V.s, in
association with the applied voltages (i.e., which is either the Bus DC voltage VDC or the root mean square rated voltage
Vrms(PW M) ) [the formulae and results are shown in the first half of Table 2 (lines 1-20)]. While the time periods t could also
be converted into angles θ via the rotational speed ω (i.e., the average current and voltage could be computed either using
angles or time intervals in the x-axis of Figure 9), the remainder of the calculation shall use time periods t. In Figure 9, the
area bound by the polygon θon O′ F ′ E ′C ′ A′ can be decomposed into 3 areas, namely that of the triangle △OC ′ A′ (= AI1 ),
the triangle △θon OA′ (= AI2 ) and the quadrilateral 2OF ′ E ′C ′ (= AI3 ). The first two are given as half the product of specific
time intervals to the corresponding current levels, and are given as
( ) ( )
1 1  
AI1 (= AOC ′ A′ ) = (tA′ C ′ )(ir ) = 1.252 × 10−3 320 = 0.2003A.s (13)
2 2

and
( ) ( )
1 1  
AI2 (= Aθon OA′ ) = (tθon A′ )(ir ) = 0.320 × 10 −3
320 = 0.512A.s (14)
2 2

The third area requires the usage of vector theory again, this time to the quadrilateral 2OF ′ E ′C ′ (in Figure 9) that holds
the vertices (t1 , i1 ) = (0, 0), (t2 , i2 ) = (tA′ C ′ + tE ′ F ′ + tC ′ E ′ , 0), (t3 , i3 ) = (tE ′ F ′ + tC ′ E ′ , i s ) and (t4 , i4 ) = (tA′ C ′ , ir ) [listed
counterclockwise along its perimeter], giving the area

tA′ C ′ + tE ′ F ′ + tC ′ E ′ tE ′ F ′ + tC ′ E ′ tE ′ F ′ + tC ′ E ′
( )
1 is
AI3 (= AOF ′ E ′ C ′ ) = + (15)
2 0 is t A′ C ′ ir

where the matrices in between the modulus are determinants, solved to be equal to the subtraction of the crossproducts.

29
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

Table 1. Computation of the EM torque and power for the SRM2 and Dual-sided SyncRM

Parameters Variable SRM2 Dual-sided Source Line

[Dimensions] SyncRM
DC bus voltage VDC [V] 500 Input 1
Number of phases q 3 Input 2
Stator (physical) poles Ns 18 Input 3
Rotor (physical) poles Nr 12 Input 4
Stator (magnetic) poles Nm 6 Input 5
Rotations per minute [rpm] 1200 Input 6
Frequency f [Hz] 60 Eq.(7) 7
Rated angular speed ω [rad/s] 125.7 Eq.(8) 8
Inductance 9
— Saturated aligned L sa [H] 0.00032 0.00043 Fig.(7) 10
— Unsaturated aligned Lua [H] 0.00644 0.00490 Fig.(7) 11
∗
— (Unsaturated aligned @i ≤ 50A) (Lua ) [H] (0.00737) (0.00543) Fig.(7) 12
— (Saturated aligned @i ≥ 120A) (L∗sa ) [H] (0.00032) (0.00043) Fig.(7) 13
— Unsaturated unaligned Luu [H] 0.00109 0.0005 Fig.(7) 14
Aligned flux-linkage intercept ΨS [V.s] 0.40456 0.35575 Fig.(7) 15
Saliency ratio 16
— (Unsaturated @i ≤ 50A) ∗
Lua /Luu 7 10 17
— (Saturated @i ≥ 120A) L∗sa /Luu 0.3 0.9 18
Rated current ir [A] 320 Input 19
Saturation current i s [A] 66.1 79.6 Eq.(5) 20
Flux-linkage @B’ Ψ(@B′ ) [V.s] 0.50717 0.49335 Fig.(7) 21
Flux-linkage @E’ Ψ(@E ′ ) [V.s] 0.425934 0.389972 Fig.(7) 22
Delta flux-linkage Ψ(@B′ ) − Ψ(@E ′ ) 0.081236 0.103378 23
[V.s] 24
Stator pole arc angle β s [deg] 10.5 Input 25
Rotation angle C → E θCE [deg] 1.17 1.49 Eq.(9) 26
Commutation factor c 0.89 0.86 Eq.(10) 27
Rms voltage Vrms(PW M) [V] 87 226 Eq.(12) 28
Energy conversion capability W ′ [J] 65 84 Eq.(6) 29
Overlap factor R 1.0476 Eq.(2) 30
Rated EM torque (recomputed) T em [N.m] 390.5 502.3 (+29%) Eq.(1) 31
(experiments* / predicted**) (415 / 390) 32
Rated EM power (recomputed) Pem [W] 49068 63126 Eq.(1) 33
(experiments* / predicted**) (50kW / 49kW) 34
*Takeno et al (2012) and **Stuikys and Sykulski (2020)

Note that the first and fourth determinant in Eq.(15) are by definition 0 due to the origin vertice being null [i.e., (x1 , y1 ) =
(0, 0)]. Expanding and re-arranging [after some simplification] gives

( ) ( )
1 1  
AI3 (= AOF ′ E ′ C ′ ) =(tE ′ F ′ + tC ′ E ′ )(ir + i s ) = 0.780 × 10−3 + 0.457 × 10−3 320 + 68.3 = 0.2402A.s (16)
2 2

In turn, the average current Iav equates to the ratio between the total sum of areas AI = AI1 + AI2 + AI3 [in A.s] and
P
the total amount of time [in seconds] taken during the trajectory path θon A C E F (these results are summarized in lines
′ ′ ′ ′

21-26 of Table 2). The same procedure (done to compute the average current) can be applied to obtain the average voltage
Vav . Finding the associated areas is simpler, as the profile of applied voltage (in blue) is square. Finally, the average
P
voltage Vav equates to the ratio between the total sum of areas AV [in V.s] and the total amount of time [in seconds].
Concluding, the Dual-sided SyncRM converts 69.3kW of average electric power into 63.1kW of electromagnetic power,
resulting in an efficiency of 91% (at the rated operational speed of 1200 rpm). This is higher than the recomputed 75%
efficiency of the SRM2 (against experiments estimation of 80% under rated torque and speed, and previous alternate
numerical prediction of 72%), yielded by the conversion of 65.3kW of average electric power into 49.1kW of EM power
(these results are summarized in lines 27-33 of Table 2).

30
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

Figure 8: Speed-torque and speed-power characteristics of the SRM2 and Dual-sided SyncRM

Figure 9: Current response as a function of voltage and rotor angular position [adapted from Stuikys and Sykulski (2020)]

31
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

5. Main Advantage
The main lesson learnt here is, that any torque dense SR motor needs to have the coils separated as much as possible
from the larger part of the ferromagnetic material (except the blocks/poles around which they are winded around), to
significantly minimize unaligned induction, thus maximizing saliency ratio. If a large chuck of ferromagnetic material
remains attached to the stator (like it is typical in conventional SR motors, where the upper ring remains attached to the
stator), during the unaligned state this will greatly conduct magnetic flux out of the coils, resulting in a large unwanted
unaligned induction response, thus significantly reducing the induction torque (and power) conversion capability of the
motor. This benefit (of decreasing the unaligned induction) by further splitting the stator (just above the coils, and
reassigning this material to the rotor) largely outweighs the detriment (of decreasing the aligned induction) of adding
another gap to the circuit. If this design change is incorporated, benefits of up to 29% in torque and power conversion can
be achieved, as well as an efficiency boost up to 91% (at the rated operational speed of 1200rpm), in concentrated-coiled
radial-flux synchronous reluctance motors.

Table 2. Efficiency computation for the SRM2 and Dual-sided SyncRM

Parameters Variable SRM2 Dual-sided Source Line

[Dimensions] SyncRM
Flux-linkage Ψuu (@A′ ) [V.s] 0.3488 0.4429 Fig. 7 1
DC bus voltage VDC [V] 500 Input 2
Time θ → A′ tθA′ [s] 0.698 × 10−3 0.320 × 10−3 Ψuu (@A′ )/VDC 3
(18%) (11%) 4
Delta flux-linkage dΨ1 [V.s] = 0.1131 0.2829 Fig. 7 5
= Ψ sa (@C ′ ) − Ψuu (@A′ ) 6
Rms voltage Vrms(PW M) [V] 87 226 Eq.(12) 7
Time A′ → C ′ tA′ C ′ [s] 1.296 × 10−3 1.252 × 10−3 dΨ1 /Vrms(PW M) 8
(34%) (45%) 9
Flux-linkage Ψua (@E ′ ) [V.s] 0.4257 0.3900 Fig. 7 10
DC bus voltage VDC [V] 500 Input 11
Time E ′ → F ′ tE ′ F ′ [s] 0.851 × 10−3 0.780 × 10−3 Ψua (@E ′ )/VDC 12
(23%) (28%) 13
Delta flux-linkages dΨ2 [V.s] = 0.081236 0.103378 Fig. 7 14
= Ψ(@C ′ ) − Ψ(@E ′ ) 15
Rms voltage Vrms(PW M) [V] 87 226 Eq.(12) 16
Time C ′ → E ′ tC ′ E ′ [s] 0.931 × 10−3 0.457 × 10−3 dΨ2 /Vrms(PW M) 17
(25%) (16%) 18
3.776 × 10−3 2.809 × 10−3
P
Total Time tT [s] t 19
(100%) (100%) 20
Area of △θon OA′ Aθon OA′ [A.s] 0.3375 0.2003 Eq.(13) 21
Area of △OA′C ′ AOA′ C ′ [A.s] 0.1116 0.0512 Eq.(14) 22
Rated current ir [A] 320 Input 23
Saturation current i s [A] 58.7 68.3 Input 24
Area of 2OF ′ E ′C ′ AOF ′ E ′ C ′ [A.s] 0.2073 0.2402 Eq.(16) 25
P
Total Area AIt [A.s] 0.6565 0.4917 A 26
Overlap factor R 1.0476 Eq.(2) 27
Average Current Iav [A] 182 183 AIt /tT 28
Average Voltage Vav [V] 358 378 AVt /tT 29
Average Electric Power Pel [kW] 65.3 69.3 Pel = Iav .Vav 30
Electromagnetic Power Pem [kW] 49.1 63.1 Eq.(1) 31
Efficiency η 75% 91% Pem /Pel 32
(experiments* / predicted**) (80% / 72%) 33
*Takeno et al (2012) and **Stuikys and Sykulski (2020)

6. Conclusion
Two ways to enhance the electromagnetic torque, power output and efficiency of a synchronous reluctance motor are:
(1) increasing the inductance of the magnetic circuits formed by the stator and rotor when aligned, or (2) reducing their
inductance when they are unaligned. Both cases result in an increase of the (unsaturated) saliency ratio Lua /Luu . This

32
http://apr.ccsenet.org Applied Physics Research Vol. 16, No. 1; 2024

article explores the second approach. It achieves this by partitioning the magnetic material of the stator, just radially
outwards from the winding, and reassigned this to the rotor. By doing so, this change further minimizes the amount of
magnetic material in contact with the coils during the unaligned case (i.e., reducing Luu ), boosting the co-energy W ′ and
ultimately imparted an increase in reluctance torque (and electromagnetic power) by 29% of an existing SRM2 reluctance
motor proposed to power the Toyota Yaris hybrid electric vehicle. Moreover, the conversion efficiency from electric to
electromagnetic power raised to 91% with a saliency ratio of 10 (values applicable at the rated operational speed of 1200
rpm). This substantial performance improvement comes at the expense of a somewhat more mechanically intense (but
achievable) machine to manufacture and assemble, with the final decision of advancing with such a modification resulting
from a careful trade-off of performance gain versus mechanical complexity (encompassing what is acceptable as part of
the project, within which this enhanced Dual-sided SyncRM is to be commercialized).
References
AKSteel. (2007). Selection of Electrical Steels for Magnetic Cores. Technical datasheet.
Arslan, B. I. (2021). Learn Drafting with NanoCAD 5.0 (ECO Edition). Amazon.com
CADrays free software. (2023). Open Cascade Company. Retrieved from https://www.opencascade.com/products/cad-
rays/
Guilera, L. C. (2018). Analytical Design of a Synchronous Reluctance Motor. Thesis: Escola Tècnica Superior d’Enginyeria
Industrial de Barcelona.
Ionel, D. M., & Popescu, M. (2011). Finite-Element Surrogate Model for Electric Machines With Revolving Field
— Application to IPM Motors. IEEE Transactions on Industry Applications, 46(6), 2424-2433. Retrieved from
https://ieeexplore.ieee.org/document/5580086
Ladislav Knebl, L., Bianchi, N., Bacco, G., & Ondrusek, C. (2020). Synchronous Reluctance Motor Analytical Model
Cross-saturation and Magnetization Analysis. Przeglad Elektrotechniczny. https://doi.org/10.15199/48.2020.01.24
Meeker D. Finite Element Method Magnetics (FEMM). FEMM Reference Manual. Retrieved from http://www.femm.info-
/wiki/Documentation
Menzies, R. W. (1972). Synchronous Reluctance Motors having Rotor Segments of Extended Pole Span. Canadian Patents
and Development Limited, U.S. Patent 3,671,789.
Misao, N., Hironori, N., & Tsunehiro, Y. (2005). High Silicon Steel Sheets Realizing Excellent High Frequency Reactor
Performance. JFE Technical Report, No. 6.
NanoCAD free software. Retrieved from https://nanocad.com/products/nanocad-free
Qi, F., Scharfenstein, D., Weiss, C., Müller, C., & Schwarzer, U. (2019). Motor Handbook. Infineon Technologies AG,
Institute for Power Electronics and Electrical Drives, RWTH Aachen University.
Shapiro, I. L., & de Berredo-Peixoto, G. (2013). Lecture notes on Newtonian mechanics: Appendix A - Fundamentals of
Vector Analysis. Springer. Retrieved from https://catalogue.library.cern/literature/e15vm-vgn59
Stuikys, A., & Sykulski, J. (2020). Average Rated Torque Calculations for Switched Reluctance Machines Based on
Vector Analysis. IntechOpen Access Books. http://dx.doi.org/10.5772/intechopen.89097
Takeno, M., Chiba, A., Hoshi, N., Ogasawara, S., Takemoto, M., & Rahman, M. A. (2012). Test results and torque
improvement of the 50-kW switched reluctance motor designed for hybrid electric vehicles. IEEE Transactions on
Industry Applications, 48(4), 1327-1334.
Wakisaka, T., Arai, S., & Kurosaki, Y. (2013). Electrical Steel Sheet for Traction Motor of Hybrid/Electrical Vehicles.
Nippon Steel Technical Report, 103, 116-120. Retrieved from https://www.nipponsteel.com/en/tech/report/nsc/pdf/103-
18.pdf
YouYou Technology Co. Ltd. (1993). JFE Super Core 10JNEX900/ 10JNHF600/ 10JNRF/ 15JNSF950. Technical data
sheet. Retrieved from https://www.10jnex900.com/

Copyrights
Copyright for this article is retained by the author(s), with first publication rights granted to the journal.
This is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license
(http://creativecommons.org/licenses/by/3.0/).

33