https://doi.org/10.3311/PPee.

19714
Creative Commons Attribution b |205
Periodica Polytechnica Electrical Engineering and Computer Science, 66(2), pp. 205–214, 2022

Parameter Determination and Drive Control Analysis of Axial

Flux Permanent Magnet Synchronous Motors

Attila Nyitrai1*, Gergely Szabó2, Sándor R. Horváth2

1
Multidisciplinary Doctoral School of Engineering Sciences, Széchenyi István Univeryity, H-9026 Győr, Egyetem tér 1., Hungary
2
Department of Electric Power Engineering, Faculty of Electrical Engineering and Informatics, Budapest University of Technology
and Economics, H-1111 Budapest, Egry József street 18., Hungary
*
Corresponding author, e-mail: [email protected]

Received: 16 December 2021, Accepted: 11 March 2022, Published online: 31 March 2022

Abstract
Axial flux electric motors have received a lot of attention in recent years due to successful implementations in industrial or traction
applications. Particularly, axial flux permanent magnet synchronous motors (AFPMSM) can be an attractive choice in case of high
torque-density requirements or when the drive environment (packaging) is geometrically limited to a disc-shaped motor. However,
compared to radial flux motors, axial flux machine modeling possibilities are much less documented. In the present study, different
electromagnetic modeling approaches have been compared through an example AFPMSM design. The motor parameters were
determined by analytical and finite element methods. A 2D equivalent model (2D Linear Motor Modeling Approach – 2D-LMMA) and
a 3D model results have been compared. The calculated values were used to carry out a drive control analysis of the axial flux motor.
Keywords
axial flux motor, motor parameters, drive control simulation, electromagnetic analysis, finite element analysis

1 Introduction
Axial flux (AF) permanent magnet synchronous motors environment. On the other hand, the output power of the AF
(AFPMSM) have some particular advantages when com- motor increases more significantly with the diameter when
pared to radial flux (RF) counterparts. An obvious advan- the active length has less influence on the performance [3].
tage compared to RF motors is the usually higher specific Applications, where high specific power (torque) is nec-
torque when using simple cooling arrangements [1, 2]. essary is therefore the area where AF motors can demon-
The electromagnetic torque of the axial flux motor is pro- strate comparative advantage to RF motors.
portional to the third power of the outer diameter of the Generally, in order to evaluate and compare the motor
active parts [3]: characteristics of different designs in real-world working
conditions, the modeling of the different physical phenom-
TEM , AF ∝ Do3 . (1)
ena in electric machines is necessary. Recently, multiph-
In case of radial flux motors, the output electromag- ysics analysis has been in the focus of many researchers,
netic torque is proportional to the rotor volume (a classi- where the advantages of such modeling approaches were
cal sizing method of RF motors is based on the torque per presented. In [4] particularly, a multiphysics modeling
rotor volume (TRV) value): approach through the design of a permanent magnet assisted
synchronous reluctance motor (PMaSynRM) were pre-
TEM , RF ∝ Dro2 L, (2)
sented. It was shown, that the importance of multiphysics
where Dro is the rotor outer diameter and L is the work- analysis in order to optimize high specific power machines
ing active length of the machine. From the above equa- is vital. The multiphysics model of the electric machine
tions it can be seen, that the axial and radial flux motors addresses at least the following physical phenomena:
have different proportions (L/D); AF motors are less suit- • Electromagnetic Analysis;
able for smaller power applications especially when the • Thermal Analysis;
outer diameter is limited by packaging constraints of the • Mechanical Analysis;

Cite this article as: Nyitrai, A., Szabó, G., Horváth, S. R. "Parameter Determination and Drive Control Analysis of Axial Flux Permanent Magnet Synchronous
Motors", Periodica Polytechnica Electrical Engineering and Computer Science, 66(2), pp. 205–214, 2022. https://doi.org/10.3311/PPee.19714
206|Nyitrai et al.
Period. Polytech. Elec. Eng. Comp. Sci., 66(2), pp. 205–214, 2022

• Structural (Strength Analysis); interior permanent magnet (IPM) or permanent magnet

• Rotordynamic Analysis; assisted synchronous reluctance machines (PMaSynRM).
• Drive Control Analysis; This category of motors has a significant reluctance
• NVH Analysis. torque component in their electromagnetic torque. From
the motor parameters' point of view, these motors are gen-
In Sections 2 and 3, as part of a multiphysics motor erally more complicated, since the two main inductance
model of the axial flux machine, an electromagnetic and parameters Ld and Lq are different. In addition, magnetic
drive control analysis with parameter determination is pre- saturation affects the two inductance components differ-
sented. In Section 1, the importance of axial flux motors ently. In many cases, due to the nonlinear behavior of the
and their research are presented. The reason for parame- magnetic model of these machines, it is more convenient
ter estimation of electric motors is discussed. The different to prepare a parameter sweep through the corresponding
approaches of parameter estimation are briefly described. Id and Iq values and determine the flux-linkages directly.
In Section 2, the electromagnetic modeling methods of These flux-linkage tables (maps) parameterized with Id
axial flux motors are presented. In addition to a simple and Iq can be directly used during numerical simulations.
analytical model, 2D and 3D finite element models are pre- However, the drawback of this method is that compared to
sented. The parameters of a given example AF motor are the motor parameter approach, the nonlinear behavior of
calculated using the different approaches and compared in the machine is rather hidden from the designer.
Subsection 2.4. Based on the calculated parameters, the
drive control analysis of the motor is presented in Section 3. 2 Electromagnetic modeling of axial flux motors
The 2D and 3D analysis was prepared using FEMM and AFPMSMs are known to be three-dimensional structures
Maxwell software respectively. The drive control analysis regarding their electromagnetic properties. The disc-
was carried out using Matlab/Simulink software. shaped airgap has a flux-density distribution that is not
only varies by the angle around the axis of rotation, but
1.1 Parameter estimation it is a function of the radial dimension as well. Therefore,
The motor parameters can be used in numerical simula- several modeling approaches were introduced [5–7] to
tions by applying system models in order to determine prepare low or high-fidelity electromagnetic models of AF
the dynamic behavior of the motor during real load-cy- machines suitable for different calculations such as sizing,
cle conditions. In case of traction motors, it is important analytical evaluation of the magnetic circuit. Generally,
to evaluate the motor performance during standardized lower fidelity 2D FEA-based methods are used to evaluate
drive-cycles in order to prove that the machine will meet nonlinear effects and naturally the highest fidelity models
the specifications of electromagnetic and thermal perfor- can be built by applying the finite element method on 3D
mance. Usually, due to the computationally expensive geometries. In [8], the authors presented a dynamic model
finite element model, it is not possible to directly couple of an AFPMSM made from soft magnetic composite core
electromagnetic and thermal field solvers with the other material. The authors used magnetic equivalent circuits to
components of the system model, such as the drive and analyze the quasi-static electromagnetic behavior. In [9],
control model. This is especially the case in early design the authors described a both steady-state and transient
iterations. Therefore, reduced order models are applied in finite element modeling of axial flux permanent magnet
order to evaluate the models in a reasonable time. In order motors and compared the results with measurements. An
to define the basic parameters of the motor, generally, the analytical non-linear magnetic circuit modeling approach
magnetic circuit model of the motor needs to be evaluated. was presented in [10], where the authors validated the
The motors' parameters can be used for further analytical model results with 3D-FEA. In [11], the authors presented
investigations in order to find working point parameters a quasi-3D magnetic equivalent circuit modeling approach
and therefore to quickly evaluate the performance of the and validated the model by measurements.
machine and drive.
In case of permanent magnet synchronous motors 2.1 Analytical methods
(PMSM), we can basically group the machines to motors The traditional approach to determine electric motor
with magnetically isotropic and anisotropic rotors. Latter working point properties is the analytical electromagnetic
machines with rotor saliency can be either classified as modeling using the lumped parameter (lumped circuit)
 |207
Nyitrai et al.
Period. Polytech. Elec. Eng. Comp. Sci., 66(2), pp. 205–214, 2022

modeling method (LPM or LCM). In this case, a magnetic

equivalent circuit model of the machines is built in order
to determine the main magnetic flux components in the
motor and therefore through the magnetic flux-linkages,
calculate the inductance values. In the present study, an
analytical approach described in [3] was implemented.
The PM flux:

 B pk  Ro2  Ri2   (3)

 PM  ,
p 2
where α is the ratio of average to peak flux densities:
Bavg
 . (4)
B pk

The armature fluxes in the d- and q-axes respectively:

2  Ro2  Ri2 (5)

 ad ,q  B pkad ,q1 .
 p 2
Fig. 1 Geometrical arrangement of the active components of an axial
From the fluxes and number of turns, the flux-linkages in flux motor. Do is the outer diameter, Di is the inner diameter, Dsn is the
the d- and q-axes can be determined respectively. diameter of the nth 2D segment used in the equivalent model.
The d- and q-axis flux-linkages therefore:
Table 1 Main parameters of the calculated motor.
1 Parameter Value Dimension
 d ,q  N1k w1 ad ,q . (6)
2 DC Voltage 670 [VDC]
Max. Current 250 [ARMS]
The d- and q-axis inductances are therefore can be calcu- Number of Poles 20 [–]
lated by the expression: Stator outer diameter (SOD) 200 [mm]

Lad ,q 
 ad ,q 1Nk 
 m0  1 w1 
2
R
2
o  Ri2 
k fd ,q . (7)
Stator inner diameter (SID) 130 [mm]

I ad ,q  p  g 'd ,q Rotor outer diameter (ROD) 200 [mm]

Rotor inner diameter (RID) 130 [mm]
The equation of the electromagnetic torque: Magnet thickness 6 [mm]
Airgap length 0.8 [mm]
3
TEM   p  m  iq   Ld  Lq   id  iq  , (8)
2
Approach) as presented in [5]. Using the multi-slice
where TEM is the electromagnetic torque, λm is the perma- method, or the 2D-LMMA, the axial flux motor can be
nent magnet flux-linkage, Ld , Lq are the d- and q-axis induc- modeled in two dimensions; since the magnetic flux-den-
tances and id , iq are the d- and q-axis currents respectively. sity distribution varies in the radial direction, depending
Equations (3)–(8) have been applied using the main on the particular design, a set of 2D equivalents, or "slices"
parameters and geometry of the motor (Fig. 1 and Table 1). can be introduced to estimate the whole motor's magnetic
The results are presented and compared to the other meth- properties with less error [12]. However, the same method
ods in Subsections 2.3 and 2.4. The analytical approach can be applied by directly solving the Maxwell-equations
considers an even distribution of the airgap flux-density on simple geometries as presented by [6]. The disadvan-
along the radial dimension of the airgap. tage of the latter method is that it is less generally applica-
ble on different geometries than the finite element method.
2.2 2D finite element methods The analysis steps of the 2D equivalent finite element
Generally, the possible 2D modeling approaches for modeling of the AF motor are the followings:
dual-rotor axial-flux motors are 2D-LMMA (Linear Motor 1. Determine the model symmetries in the direction of
Modeling Approach), 2D-IRMA (Internal Rotor Modeling rotation based on the number of slots ( NS ) and num-
Approach) and 2D-ORMA (Outer Rotor Modeling ber of poles (2P);
208|Nyitrai et al.
Period. Polytech. Elec. Eng. Comp. Sci., 66(2), pp. 205–214, 2022

2. Determine the number of slices and prepare the 2D

geometric representations of the motors;
3. Application of the electromagnetic boundary condi-
tions of the Linear Motor Modeling Approach as pre-
sented in Fig. 2;
4. Definition of the magnetic material properties;
5. Application of the electromagnetic excitations;
6. Meshing and mesh sensitivity analysis;
7. Solution of the 2D finite element models;
8. Evaluation of the results.

The results of the 2D FEA are presented in Fig. 3 and

Fig. 4. In Fig. 3, the magnetic flux density plot is pre-
Fig. 4 Permanent magnet flux linkages of the different slices
sented. In Fig. 4, the flux-linkages of the different sections of the 2D equivalent model. Slice 1: λ1, slice 2: λ2 slice 3: λ3
can be observed.
for modeling anisotropic, three-dimensional structures
2.3 3D finite element methods without an easy-to-prove 2D equivalent, such as in the
Compared to the 2D approach using the multi-slice method, case of axial flux machines.
the 3D finite element method ensures a proper modeling of An additional benefit of using 3D analysis is the pos-
a set of three-dimensional effects including the radial air- sibility for proper consideration of end-effects, offer-
gap flux density distribution of AF motors. Obviously, 3D ing more precise calculation of e.g. end-winding leakage
electromagnetic modeling is a straightforward approach and related stray losses in structural parts. Such end-ef-
fects are to be investigated dominantly in radial direction
regarding dual rotor, sandwich-like axial flux topologies.

2.3.1 Yokeless and segmented stator modeling

Generally, in terms of rotating electric machines, a yoke is
considered as a material of high magnetic permeability for
providing a flux path between adjacent poles.
Considering technological possibilities of winding-au-
tomation, the stator core should be segmented to obtain a
feasible geometry for mass production. Such yokeless sta-
tor structure can be observed in Fig. 5, where each coil is
to be wound around an individual segment.
Fig. 2 A slice of the 2D equivalent model of the motor.
On the model boundaries periodic and A=0 boundary
conditions should be applied.

Fig. 3 Magnetic flux density plot and flux lines of the equivalent 2D Fig. 5 Yokeless and segmented stator design with simplified winding
model in open-circuit condition. geometry, prepared for 3D electromagnetic analysis.
 Nyitrai et al.
Period. Polytech. Elec. Eng. Comp. Sci., 66(2), pp. 205–214, 2022 |209

The main advantages of applying this yokeless stator

design are related to mass reduction, (increased specific
power [kW/kg]) and efficiency improvement due to lower
total core loss gained by the elimination of hysteresis loss.

2.3.2 Rotating band setup

In order to model the relative displacement between sta-
tionary and rotating parts, most software require a special
region, usually called band, to be defined at the centerline
of airgap. In case of an axial-flux PMSM with dual rotor
Fig. 7 Mesh of the 3D model. The rotor and stator represented
topology, as illustrated by Fig. 6, the practical approach is
separately.
to define a single band in such a way that it encloses all the
stator core segments and windings. By applying a rotating Table 2 Comparison of the mesh properties.
band, the need for complete remeshing at every time step Variable name 2D FEA 3D FEA
can be avoided, leading to a significant reduction of simu- Element type Triangle Tetrahedron
lation time and computational resources. Number of elements 45203 243199

2.3.3 Mesh generation at 20 [°C]. The simulated no-load line-to-line voltage at

Discretization of the 3D model space is carried out using each time step over a complete electrical period is pre-
tetrahedron elements. During a transient solution pro- sented in Fig. 8. Applying previously presented mesh den-
cess, adaptive mesh generation (typical e.g. in case of sity for rough transient study, the calculated average torque
eddy-current problems) is not possible, so the relevant set is practically zero, with numerical pulsation between
of Maxwell's equations are evaluated for the same nodes ±200 [mNm] region.
at every time step. Proposed initial mesh density for rough
3D transient analysis is presented in Fig. 7. 2.3.5 Simulation of short-circuit characteristics
In volumes where the intensity of magnetic field is Just like the similar characteristic measurement, the simula-
high, usually around the air gap and end-winding area of tion of short-circuit condition should be performed in gen-
a rotating machine, mesh quality holds key importance for erator mode, forcing a constant shaft speed regardless the
simulation fidelity and time demand of calculations. magnitude of electromagnetic torque and output current.
The mesh type and the number of the elements is indi- Short-circuit characteristics obtained by transient 3D
cated in Table 2 in case of 2D and 3D FEA, corresponding simulation are presented in below Fig. 9.
to the precision level of presented results. According to IEEE 812-1984 standard [13], performing
short-circuit measurement with a suitable external reac-
2.3.4 Simulation of no-load characteristics tance (limiting the current e.g. to its rated value) is a rea-
The peak value of the flux-linkage is 0.064 [Vs] using the sonable approach to obtain steady-state temperature rise
selected magnet grade N35H ( Hc = −840 [kA/m], μr = 1.06) when inverter-coupled test cannot be carried out.

Fig. 6 3D model and rotating band setup of the yokeless axial flux
permanent magnet motor topology. Fig. 8 No-load Line-to-Line voltage at 2200 [1/min].
210|Nyitrai et al.
Period. Polytech. Elec. Eng. Comp. Sci., 66(2), pp. 205–214, 2022

2.3.6 Simulation of rated load

The electromagnetic (airgap) torque of synchronous
machines is traditionally expressed as a function of
load-angle δ, which is expected to be sinusoidal in case of
a surface-mounted (magnetically symmetric) rotor design.
Flux-density in magnetically active parts, corresponding to
maximum torque operating point, is visualized by Fig. 10.
Based on identical operating conditions, the compari-
son of torque profiles between 3D FEA and 2D-LMMA
methods can be studied in Fig. 11.
In Fig. 10 linear scale was applied between 0 [T] – 2.2 [T]
to visualize magnetic field. Flux-density in Permanent
Fig. 9 Short-circuit current at 40 [1/min] speed.
Magnet parts was also checked to ensure that rated operat-
ing point is feasible with reversible demagnetization.

Fig. 10 Flux-density of stator segments at rated load condition.

 |211
Nyitrai et al.
Period. Polytech. Elec. Eng. Comp. Sci., 66(2), pp. 205–214, 2022

Result for each load-angle was obtained by 3D mag- In Table 3, the motor parameters are compared. The
netostatic simulation, using a self-developed scripting analytical torque value calculated from Eq. (8) in case of
method to eliminate repetitive human procedures. q-axis current is 186 Nm, therefore the error compared to
the result from the 3D model is 11%.
2.4 Comparison of the results of the different methods
The calculated motor parameter values have been com- 3 Drive control analysis
pared. The d- and q-axis inductances and the PM flux-link- 3.1 Dynamic concentrated parameter model
ages have been compared applying the different calcula- The drive control part of the analysis attempted to test the
tion methods. From the results the following conclusions machine's behavior under controlled dynamic conditions.
have been drawn. In case of an early design analysis, the This approach is based on the machine's differential sys-
analytical method is appropriate. On the other hand, the tem of equation which is the following for the selected
simple analytical calculation does not include non-linear machine type (AFPMSM):
effects, such as magnetic saturation. Regarding the FEA did
ud  Rid  Ld   p Lq iq , (9)
methods, the 2D equivalent and the 3D FEA results are
dt
suitable for the further drive control analysis, because the
difference was approximately 12.5%. diq
uq  Riq  Lq   p Ld id   p  p , (10)
It was found, that the computation time of the 3D model dt
was significantly longer, however, the multi-slice method
 
3
required more manual work to prepare the geometry of m p  Ld  Lq  id iq   p iq , (11)
2
the sections. If the geometry preparation was automatized,
d (12)
then it can be concluded, that the 2D FEA method is pre- J  m  ml  F  ,
dt
ferred, in case of geometrical optimization and when more
variants are also necessary to be evaluated. The authors where ud, uq are the real and imaginary parts of the stator
propose a workflow where the motor parameters are cal- voltage vector, id, iq are the real and imaginary parts of the
culated in the following order: stator current vector respectively, F is the friction loss fac-
Pre-design by analytical motor parameter calculation tor, J is the moment of inertia, ω is shaft angular speed.
(no saturation effects considered): The presented machine model points out that an axial
• 2D equivalent model for geometrical optimization flux, permanent magnet machine's model does not differ
and motor parameter determination in each step, from the radial one's.
considering non-linear effects; The machine model was parameterized using the pri-
• 3D FEA for the final determination of the motor ory magnetostatic FEM simulations, where inductance
parameters, considering non-linear effects on a true and flux parameters were calculated. This also defines an
3D model. improvement option [14], which can provide significantly
better, but yet computationally not too expensive machine
model, and its steps can be summarized as follows:
• calculating the flux profiles of the machines in direct
and quadratic direction using predefined current
combinations. By doing this the inductance profiles
of the machine can be calculated and its integration
into the machine model results in saturating behavior,

Table 3 Comparison of the results.

Variable name Analytical 2D FEA 3D FEA
d-axis inductance (Ld ),
218 204 208
[μH]
q-axis inductance (Lq ),
207 203 197
[μH]
PM flux-linkage (λPM ),
Fig. 11 Comparison of Torque characteristics using 0.062 0.056 0.064
[Wb.turns]
100 [Apk] supply.
212|Nyitrai et al.
Period. Polytech. Elec. Eng. Comp. Sci., 66(2), pp. 205–214, 2022

• calculating cogging torque profile of the machine.

This can be modelled as the difference of machine
model's torque and FEM calculated torque.

The developed control structure was the widely used

PI cascade control loop as shown in Fig. 12, where every
C denotes a controller and their subscripts indicate their (a)
purpose.
To achieve better utilization of the machine MTPA
algorithm was implemented [15], which provided current
reference for both current controllers. The PI-type control-
lers were modelled as follows:

 1 
GPI  s   Ap 1  , (13)
 sTi  (b)

where Ap is the proportional gain, Ti is the integral time.

The tuning was performed based on the predefined cut-
off frequency and damping factor [16], which resulted the
parameters in Table 4.

3.2 Simulation results

The simulation target was to verify the machine model's
(c)
response during no-load and nominal load conditions. The
test speed – torque covered the four quadrants.
Fig. 13 summarizes the simulation results. where the
nominal speed and load was examined. Fig. 13(a) and (b)
shows the d- and q-direction current components.

(d)

Fig. 13 Drive control analysis results; (a) d-direction current;

(b) q-direction current ; (c) angular speed; (d) torque.

The first one usually tends to have nearly zero value

expect the load transients. Beside the implemented MTPA
Fig. 12 Cascade control loop. the requested d-direction current is still relatively small,
since the direct and quadratic inductance difference is very
Table 4 Field oriented cascade control loop parameters.
small. The q-direction current component correlates with
Parameter Value Description
the load and dynamic events, and also the current control-
Apω 2.8252 Speed controller proportional gain
lers were stable. Fig. 13(c) and (d) show the mechanical
Tiω 66.7 ms Speed controller integral time
response of the motor. The angular speed control was sta-
Direct direction current controller
Apd 0.2939
proportional gain ble, with good dynamics even in load drops.
Direct direction current controller
Tid 1.1 ms
integral time 3.3 The control algorithm's further possibilities
Apq 0.4385
Quadratic direction current controller The presented control loop contains PI type control-
proportional gain
lers, but other control structures are still promising.
Quadratic direction current controller
Tiq 1.5 ms
integral time
Model Predictive Control is being published for many
 |213
Nyitrai et al.
Period. Polytech. Elec. Eng. Comp. Sci., 66(2), pp. 205–214, 2022

applications, including field-oriented controls of rotating AFPM Axial flux permanent magnet (~motor);
machines [17, 18]. Other possibilities could be the imple- BC boundary condition;
mentation of sensorless algorithms, which offers many EMAG electromagnetic;
possibilities. EMF electro-motive force;
One approach could be a model-based solution, for FE finite element;
example an Extended-Kalman filter-based approach [19]. FEA finite element analysis;
Another solution could be a non-model-based solution, FEM finite element modeling;
such as high-frequency voltage injection methods. [20, 21]. IRMA Internal Rotor Modeling Approach;
LMMA Linear Motor Modeling Approach;
4 Conclusion MEC magnetic equivalent circuit;
In the present study, different methods for the determina- PM permanent magnet;
tion of axial flux motor parameters have been presented. RF radial flux;
Analytical, 2D and 3D finite element methods have been RFPM radial flux permanent magnet (~motor);
compared in case of an example AFPMSM. It was found, SPM surface (-mounted) permanent magnet;
that analytical methods for the d- and q-axis inductance SynRM synchronous reluctance motor.
calculation are suitable for the quick estimation of the
motor parameters, however, in case of complicated geom- List of symbols
etries, and in order to consider non-linear behaviour, a ΦPM permanent magnet flux;
general numerical method is more practical to use. The 2D α ratio of average and peak ;
finite element model, compared to the 3D one was found Bavg average flux-density;
to reduce the computational time significantly, however, Bpk peak flux-density;
more manual work and more complicated model setup is Ro outer radius;
necessary because of the equivalent geometries. During Ri inner radius;
the drive control analysis, it was found that similar meth- Ψd,q flux-linkage (d or q axis);
ods used in case of radial flux PMSMs were sufficient. N1 number of turns per phase;
Using the FEA calculated motor parameters the built-up k w1 winding factor (fundamental);
drive control simulation was able to reproduce the nomi- Φad,q armature flux (d or q axis);
nal load operating of the motor, and provided stable con- Lad,q boundary condition;
trol in the four quadrants. Authors are going to conduct TEM electromagnetic torque;
further research in order to establish integrated design p number of pole pairs;
techniques for AFPMSM machines. λm magnet flux-linkage;
iq q-axis current;
Nomenclature id d-axis current;
AF axial flux; Ld d-axis inductance;
AFIR axial flux internal rotor; Lq q-axis inductance.

References
[1] Patterson, D. J., Colton, J. L., Mularcik, B., Kennedy, B. J., [3] Gieras, J. F., Wang, R. J., & Kamper, M. J. " Principles of AFPM
Camilleri, S., Rohoza, R. "A comparison of radial and axial flux Machines", In: Axial Flux Permanent Magnet Brushless Machines,
structures in electrical machines", In: 2009 IEEE International Springer, Dordrecht, Netherlands, 2005, pp. 27–78.
Electric Machines and Drives Conference, Miami, FL, USA, 2009, https://doi.org/10.1007/1-4020-2720-6_2
pp. 1029–1035. [4] Nyitrai, A., Szabó, G., Horváth, S., Vajda, I. "Multiphysics
https://doi.org/10.1109/IEMDC.2009.5075331 Analysis of Automotive PMaSynRM", In: 2019 International
[2] Cavagnino, A., Lazzari, M., Profumo, F., Tenconi, A. "A compar- IEEE Conference and Workshop in Óbuda on Electrical and Power
ison between the axial flux and the radial flux structures for PM Engineering (CANDO-EPE), Budapest, Hungary, 2019, pp. 67–72.
synchronous motors", IEEE Transactions on Industry Applications, https://doi.org/10.1109/CANDO-EPE47959.2019.9111009
38(6), pp. 1517–1524, 2002. [5] Gulec, M., Aydin, M. "Implementation of different 2D finite element
https://doi.org/10.1109/TIA.2002.805572 modelling approaches in axial flux permanent magnet disc machines",
IET Electric Power Application, 12(2), pp. 195–202, 2018.
https://doi.org/10.1049/iet-epa.2017.0434
214|Nyitrai et al.
Period. Polytech. Elec. Eng. Comp. Sci., 66(2), pp. 205–214, 2022

[6] Tiegna, H., Bellara, A., Amara, Y., Barakat, G. "Analytical [14] Szabó, G., Veszprémi, K. "FEM-Parameterized Sensorless Vector
Modeling of the Open-Circuit Magnetic Field in Axial Flux Control of PMSM Using High-Frequency Voltage Injection", In:
Permanent-Magnet Machines with Semi-Closed Slots", IEEE Doctoral Conference on Computing, Electrical and Industrial
Transactions on Magnetics, 48(3), pp. 1212–1226, 2012. Systems, Costa de Caparica, Portugal, 2021, pp. 226–239.
https://doi.org/10.1109/TMAG.2011.2171979 https://doi.org/10.1007/978-3-030-78288-7_22
[7] Egea, A., Almandoz, G., Poza, J., Ugalde, G., Escalada, A. J. [15] Preindl, M., Bolognani, S. "Optimal State Reference Computation
"Axial-Flux-Machine Modeling With the Combination of FEM- With Constrained MTPA Criterion for PM Motor Drives", IEEE
2-D and Analytical Tools", IEEE Transactions on Industry Transactions on Power Electronics, 30(8), pp. 4524–4535, 2015.
Applications, 48(4), pp. 1318–1326, 2012. https://doi.org/10.1109/TPEL.2014.2354299
https://doi.org/10.1109/TIA.2012.2199450 [16] Wang, L., Chai, S., Yoo, D., Gan, L., Ng, K. "PID and Predictive
[8] Maloberti, O., Figueredo, R., Marchand, C., Choua, Y., Control of Electrical Drives and Power Converters using
Condamin, D., Bommé, E. "3-D–2-D Dynamic Magnetic Modeling MATLAB/Simulink", John Wiley & Sons Inc., Singapore,
of an Axial Flux Permanent Magnet Motor With Soft Magnetic Singapore, 2015.
Composites for Hybrid Electric Vehicles", IEEE Transactions on [17] Morel, F., Lin-Shi, X., Retif, J. M., Allard, B., Buttay, C. "A
Magnetics, 50(6), Article number: 8201511, 2014. Comparative Study of Predictive Current Control Schemes
https://doi.org/10.1109/TMAG.2014.2300152 for a Permanent-Magnet Synchronous Machine Drive", IEEE
[9] Hill-Cottingham, R. J., Coles, P. C., Rodger, D., Lai, H. C. "Finite Transactions on Industrial Electronics, 56(7), pp. 2715–2728, 2009.
element modelling of an axial flux PM machine", In: Second https://doi.org/10.1109/TIE.2009.2018429
International Conference on Power Electronics, Machines and [18] Preindl, M., Bolognani, S. "Model Predictive Direct Speed
Drives (PEMD 2004), Edinburgh, UK, 2004, pp. 552–555. Control with Finite Control Set of PMSM Drive Systems", IEEE
https://doi.org/10.1049/cp:20040347 Transactions on Power Electronics, 28(2), pp. 1007–1015, 2013.
[10] Kano, Y., Kosaka, T., Matsui, N. "A Simple Nonlinear Magnetic https://doi.org/10.1109/TPEL.2012.2204277
Analysis for Axial-Flux Permanent-Magnet Machines", IEEE [19] Dilys, J., Stankevič, V., Łuksza, K. "Implementation of Extended
Transactions on Industrial Electronics, 57(6), pp. 2124–2133, 2010. Kalman Filter with Optimized Execution Time for Sensorless
https://doi.org/10.1109/TIE.2009.2034685 Control of a PMSM Using ARM Cortex-M3 Microcontroller",
[11] Vansompel, H., Sergeant, P., Dupré, L. "Optimized Design Energies, 14(12), Article number: 3491, 2021.
Considering the Mass Influence of an Axial Flux Permanent-Magnet https://doi.org/10.3390/en14123491
Synchronous Generator With Concentrated Pole Windings", IEEE [20] Szabó, G., Veszprémi, K. "Sensorless Vector Control of Permanent
Transactions on Magnetics, 46(12), pp. 4101–4107, 2010. Magnet Synchronous Machine Using High-Frequency Signal
https://doi.org/10.1109/TMAG.2010.2070075 Injection", Acta Polytechnica Hungarica, 17(4), pp. 145–163, 2020.
[12] Nyitrai, A., Orosz, T. "FEM-Based Benchmark Problem for https://doi.org/10.12700%2FAPH.17.4.2020.4.8
Cogging Torque Minimization of Axial Flux Permanent-Magnet [21] Szabó, G., Veszprémi, K. "Sensorless Vector Control of Permanent
Motors in Artap Framework", Periodica Polytechnica Electrical Magnet Synchronous Machine Using High-Frequency Rotating
Engineering and Computer Science, 65(2), pp. 152–159, 2021. Injection", In: 2021 IEEE 19th International Power Electronics
https://doi.org/10.3311/PPee.17755 and Motion Control Conference (PEMC), Gliwice, Poland, 2021,
[13] Institute of Electrical and Electronics Engineers "IEEE 812-1984 pp. 588–593.
IEEE Standard Definitions of Terms Relating to Fiber Optics", https://doi.org/10.1109%2FPEMC48073.2021.9432563
Institute of Electrical and Electronics Engineers, Piscataway, NJ,
USA, 1984.