energies

Article
Modeling of Axial Flux Permanent Magnet Generators
Natalia Radwan-Pragłowska, Tomasz W˛egiel * and Dariusz Borkowski
Faculty of Electrical and Computer Engineering, Cracow University of Technology, Warszawska 24 St.,
31-155 Cracow, Poland; natalia.radwan-praglowska@pk.edu.pl (N.R.-P.); dborkowski@pk.edu.pl (D.B.)
* Correspondence: pewegiel@cyfronet.pl




Received: 28 August 2020; Accepted: 29 October 2020; Published: 2 November 2020


Abstract: This article focuses on modeling of an Axial Flux Permanent Magnet Generator (AFPMG).
The authors analyzed selected variants of disk generators, including coreless stator constructions and
with iron core ones, also taking into account the Permanent Magnet (PM) arrangement in order to
show the way to obtain the optimal machine characteristics based on analytical equations. In addition
to the full model, which takes into account the higher harmonics of the magnetic field distribution,
the paper presents a simplified mathematical model developed for generator operation cases such as
standalone, connected to a 3-phase power grid and loaded with a diode rectifier. The analytical and
finite-element method (FEM) calculations were performed as well as laboratory tests to confirm the
correctness of presented model assumptions.

Keywords: permanent magnet; PM generator; axial flux machine; disk generator

1. Introduction
The interest in synchronous generators excited by Permanent Magnets (PM) is related to the global
tendencies to support the power industry with the help of local power plants based on renewable
energy sources such as water or wind energy, in which PM synchronous generators are often used.
For these solutions, it is possible to develop a structure with a very large pole number and relatively
small dimensions. The aim of the work is related to the search for design solutions for electromechanical
energy converters with permanent magnets dedicated to small power plants. These include machines
excited by PM with a radial magnetic field as well as disk machines with an axial flux permanent
magnetic field [1–7].
The first Axial Flux Permanent Magnets (AFPM) machines appeared in the 1830s [8]. In 1831 M.
Faraday constructed a unipolar generator (the so-called Faraday disk), which was a machine with
axial magnetization. This type of machine also appeared in N. Tesla’s patent [9], published in 1889.
Earlier, in 1837, there was also a patent by T. Davenport [8], in which the first machine with axial
magnetization was described. However, the wider development of axially magnetized machines
started after 1983, when modern high-energy rare earth permanent magnets based on the combination
of Neodymium, Iron and Boron appeared.
Currently produced AFPM machines are characterized by a simple design and relatively small
dimensions, especially in the rotor axis, as well high torque-to-weight ratio and efficiency. These benefits
sometimes give AFPM machines advantages over conventional machines. AFPM machines are
commonly used in various applications. In papers from the recent years, apart from the structural
analyses of AFPM generators used for wind farms [1,3–5], there are more and more other possibilities
of their use. One can mention here electric cars, traction drives and low-speed vehicles [2,7,8,10].
The stator construction of AFPM machines can be with or without armature slots, with or without
armature core [8,10]. Among the commonly used stator designs, there are two variants of the stator
coil arrangement: non-overlapping and overlapping windings [11,12]. The structure of the rotor is

Energies 2020, 13, 5741; doi:10.3390/en13215741 www.mdpi.com/journal/energies

Energies 2020, 13, 5741 2 of 28
Energies 2020, 13, x FOR PEER REVIEW 2 of 29

based on theThe selection

stator construction
of its type:ofinternal
AFPM or machines
externalcan be with or without
permanent-magnet armature
rotors, withslots, with or
surface-mounted
without
or interior armature core [8,10].and
permanent-magnet Among the commonly
as single-stage or used stator designs,
multistage [8]. there are two variants of the
stator
In coil years,
recent arrangement:
a lot ofnon-overlapping
research has been and overlapping
carried outwindings
to optimize[11,12].
theThe structure
design of the
and modelling
rotor is based on the selection of its type: internal or external permanent-magnet rotors, with surface-
of AFPM machines [13–28]. There are considerations for which analyses are carried out strictly on
mounted or interior permanent-magnet and as single-stage or multistage [8].
the basis of calculations of the magnetic field distribution with the use of numerical methods based
In recent years, a lot of research has been carried out to optimize the design and modelling of
on the finite-element
AFPM method
machines [13–28]. There(FEM) [18,21,24] and
are considerations forwith
whichthe use ofare
analyses analytical
carried out models
strictlywhich
on the also
allowbasis
oneoftocalculations
obtain quantitatively satisfactory results. However, numerical
of the magnetic field distribution with the use of numerical methods based onmodeling techniques
do nottheeasily allow a method
finite-element synthesized(FEM)study for phenomena
[18,21,24] and with the use occurring in themodels
of analytical machine,
which hence analytical
also allow
modelsoneare very popular.
to obtain For AFPM
quantitatively machines,
satisfactory results.there are some
However, areasmodeling
numerical for whichtechniques
one can make do notsome
easily allow
interesting a synthesized
considerations study for complementation
regarding phenomena occurring of in
thetheanalytical
machine, hence analytical
models, similar models
to classic
are very popular. For AFPM machines, there are some areas for which one
aspects of electrical machine modelling. Basically, analytical models are based on two approaches, can make some interesting
considerations regarding complementation of the analytical models, similar to classic aspects of
analytical resolutions of the Maxwell’s equations [13–15,17,19,20,23] and the magnetomotive force
electrical machine modelling. Basically, analytical models are based on two approaches, analytical
(MMF) by the permeance product [16,22,25–28]. The first analytical approach leads to very complex
resolutions of the Maxwell’s equations [13–15,17,19,20,23] and the magnetomotive force (MMF) by
equations that are often difficult to accept by engineers. Mathematical models based on permeance
the permeance product [16,22,25–28]. The first analytical approach leads to very complex equations
functions are moredifficult
that are often understandable
to accept by forengineers.
designers. Additionally,
Mathematical these
models models
based allow for analyzing
on permeance functions the
formation
are moreandunderstandable
reduction of cogging torques
for designers. for iron core
Additionally, theseAFPM
modelsmachines [25–28]. the formation
allow for analyzing
The aim of this
and reduction article torques
of cogging is the presentation
for iron core AFPM of a universal analytical methodology that allows
machines [25–28].
The aim of this
for comprehensive article isof
modelling the presentation
various designs of aofuniversal
AFPM analytical
machines. methodology
This approach that allows
has notfor been
comprehensive modelling of various designs of AFPM machines. This approach
widely published for this class of machines. This paper focuses in particular on a representative has not been widely
published
case of AFPM disc for this class of machines.
generators This paper
with a stationary focuses
stator withinnon-overlapping
particular on a representative
windings and case
twoof rotor
AFPM disc generators with a stationary stator with non-overlapping windings and two rotor discs.
discs. These solutions are well known; however, the aim of this work is to modify the structure
These solutions are well known; however, the aim of this work is to modify the structure and
and consequently, to correct the analytical mathematical models. The mentioned structure was
consequently, to correct the analytical mathematical models. The mentioned structure was chosen
chosen because of its simplicity, with the possibility of modifying the stator (coreless; with cores)
because of its simplicity, with the possibility of modifying the stator (coreless; with cores) and the
and the arrangement
arrangement of theof the permanent
permanent magnets
magnets (simple;
(simple; skewed),
skewed), in order
in order to improve
to improve the generator
the generator
characteristics andand
characteristics parameters,
parameters, ininparticular
particular toto increase theobtained
increase the obtained power
power andand reduce
reduce the cogging
the cogging
torque. In this article, the four structures of AFPM generators ware considered,
torque. In this article, the four structures of AFPM generators ware considered, taking into account taking into account the
highertheharmonics
higher harmonics of flux density
of flux density distributions.
distributions. An additional
An additional aspectaspect of created
of created models models is their
is their usability
usability forpurposes,
for operational operational purposes,
e.g., e.g., cooperation
cooperation with a six-pulsewith adiode
six-pulse diodewhich
rectifier, rectifier,
was which
alsowas also
presented in
presented in this paper. The correctness of the result was obtained on
this paper. The correctness of the result was obtained on the basis of the created mathematical models,the basis of the created
and itmathematical
was verifiedmodels,
on theand basisit was verified on the basis of laboratory tests and FEM calculations.
of laboratory tests and FEM calculations.
2. Mathematical Model
2. Mathematical Model
A generator with non-overlapping stator windings and permanent magnets placed on one side
A generator with non-overlapping stator windings and permanent magnets placed on one side of
of two rotor disks was selected for the further analysis. The model created in ANSYS Maxwell
two rotor disks
software is was selected
shown for the
in Figure further
1. The analysis.
generator modelThe
wasmodel created
divided in ANSYS
into ten regions, Maxwell software is
with a separate
shown in Figure 1. The generator model was divided into ten regions, with a separate mesh
mesh defined for each. The total number of tetrahedral elements (for the basic model mesh) was defined
for each. The total number of tetrahedral elements (for the basic model mesh) was 857,643.
857,643.

(a) (b) (c)

Figure 1. Axial Flux Permanent Magnet (AFPM) generator model with ANSYS Maxwell program:
(a) stator disc with non-overlapping windings; (b) a rotor disc with basic (simple) Permanent Magnet
(PM) arrangement; (c) assembling.
 Energies 2020, 13, x FOR PEER REVIEW 3 of 29

Figure 1. Axial Flux Permanent Magnet (AFPM) generator model with ANSYS Maxwell program: (a)
stator disc with non-overlapping windings; (b) a rotor disc with basic (simple) Permanent Magnet
Energies 2020,(PM)
13, 5741
arrangement; (c) assembling. 3 of 28

The main features of considered AFPM machine with non-overlapping windings [29–32]:
The main
number features
of coils of considered
per phase AFPM
p s , total number machine
of stator coils 3 with non-overlapping
p s , number of magnets per windings
1 rotor disc [29–32]:
number of coils per phase ps , total number of stator 2π coils4π3 ps , number of magnets per 1 rotor
2p = 4p , p / ps = 2, maximum angle of coil pitch ε max  
disc 2p = 4ps s , p / p s = 2, maximum angle of coil pitch 3p εmax = .3p
3p
2π
= 4π
3p .
s s
Further
Further analyses were carried out for four generator variants: two statorstator
analyses were carried out for four generator variants: two designs
designs [30–32][30–32]
and twoand two
ways of magnets arrangement. In case of amateur solutions, these design options
ways of magnets arrangement. In case of amateur solutions, these design options are quite simple are quite simple and
and often
often performed.performed.
The authors
The authors considered
considered statortopologies
stator topologies shown
shown in in
Figure
Figure 2: a 2:
coreless statorstator
a coreless and a stator
and awith
stator with
iron cores placed inside the coils. Cases of simple and skewed arrangement of PM on the rotor disks
iron cores placed inside the coils. Cases of simple and skewed arrangement of PM on the rotor disks
were considered.
were considered.
Ri Ri
Ro Ro

Rotor Rotor
core core

PM S N lm PM S N lm
Coreless stator Stator with cores
l li l

PM N S lm PM N S lm

Rotor Rotor
core core

(a) (b)
Figure
Figure 2. Design
2. Design of of
ananAFPM
AFPM generator:
generator: (a) coreless
(a) stator;
coreless (b) stator
stator; with iron
(b) stator withcores.
iron cores.

2.1. Distribution
2.1. Distribution of Magnetic
of Magnetic Field
Field
2.1.1. Basic
2.1.1. Basic ModelModel of Flux-DensityDistribution
of Flux-Density Distribution Induced
Inducedbyby
PMPM
First of all, as the basic one for the further considerations, only the distribution of the magnetic
First of all, as the basic one for the further considerations, only the distribution of the magnetic
field from permanent magnets was used. The authors assumed a linear approximation of the
field from permanent
demagnetization magnets was
characteristics used. The
of permanent authors
magnets Bm assumed
Br  0  rm Ham linear approximation
and neglected the iron of the
demagnetization characteristics of permanent magnets
saturations and magnetic voltage drops. For the discussed cases, Bm = B + µ H
0 rm m are justified due the iron
µ
r assumptions
these and neglected
saturations and magnetic
to the relatively voltage drops.
large dimensions of theForair the
gaps.discussed cases,
Despite the these
assumed assumptions
simplicity, are justified
the problem of due
to theanalytical
relatively large dimensions
modelling of the
of the magnetic airdistribution
field gaps. Despite
becomesthemore
assumed simplicity,
complicated the problem of
and relatively
moremodelling
analytical difficult to describe than in the
of the magnetic casedistribution
field of classic machines.
becomes Based
moreon complicated
the principles and
presented in
relatively more
exemplary literature [8,13,15,29,33–35], analytical equations describing the flux-density
difficult to describe than in the case of classic machines. Based on the principles presented in exemplarydistribution
in the air gap for the base model of a coreless generator were used. The cross-section is presented in
literature [8,13,15,29,33–35], analytical equations describing the flux-density distribution in the air gap
Figures 3 and 4.
for thewhere:
base model of a coreless generator were used. The cross-section is presented in Figures 3 and 4.
where:
θ—it is the angular coordinate associated with the air-gap,
θ—it is
φ—it
theisangular
the anglecoordinate
of the rotor associated
position with respect
with thetoair-gap,
the reference frame.
ϕ—it is the angle of the rotor
Approximation of the position with respect
axial component
Energies 2020, 13, x FOR PEER REVIEW
to theflux-density
of magnetic reference frame.
distribution induced by4 of
the29
permanent magnets in the middle of the air gap (component z  0 ) for coreless AFPM machine is
presented in Figure 5 [29–32].
rotor 1
2β(r)

S PM N S N lm
 
z
0z  l
0θ

m N S N S lm

rotor 2

Basic
Figure 3.Figure model
3. Basic to to
model illustrate themagnetic
illustrate the magneticfield field distribution
distribution from permanent
from permanent magnets in themagnets
air gap. in the
air gap.
 p

N
Ro
 0θ 

m S
N PM N
S S N N llmm
  S
z
rotor 2 0z  l
0θ
Energies 2020, 13, 5741 4 of 28
m N S N S lm
Figure 3. Basic model to illustrate the magnetic field distribution from permanent magnets in the air gap.
rotor 2
 p

Figure 3. Basic model to illustrate the magnetic field distribution from permanent magnets in the air gap.
N
 p N
Ro

N N
Ri
N r
Ri N Ro
Ro
N

N N
Ri
r
(a) Ri (b) Ro
N
Figure
Figure 4. (a,b)
4. (a,b) Arrangementofofpermanent
Arrangement permanent magnets
magnetswith
withanan
exemplary shape.
exemplary shape.

Approximation of the axial component of magnetic flux-density distribution induced by the

m (   , r ) z = 0) for coreless AFPM machine is
(a) (b)
permanent magnets in the middle of the air gapB(component
presented in FigureFigure 4. (a,b) Arrangement of permanent
5 [29–32]. 2D magnets with an exemplary shape.
1D

Bm (   , r )  
 π  π
2p
  (r ) 0 2D  (r ) π π
p 1D 2p p

 
 πp  2pπ  (r ) 0  (r ) π
2p
π
Figure 5. Approximated distribution function of magnetic flux-density from ppermanent magnets in
the air gap according to    ; Models: 1D–solid line, 2D–dashed line.

The distribution of magnetic flux-density in the gap induced by permanent magnets can be
Figure 5. Approximated distribution function of magnetic flux-density from permanent magnets in the
approximated using the Fourier
Figure 5. Approximated series function
distribution [13,15,29–32]:
of magnetic flux-density from permanent magnets in
air gap according to θ − ϕ; Models: 1D—solid line, 2D—dashed line.
the air gap according to    ; Models: 1D–solid line, 2D–dashed line.
Bm (   , r ) in the
The distribution of magnetic flux-density 
j (    )
(r )  einduced
BPMgap by permanent magnets(1)can be
The distribution of magnetic flux-density in  Qthe gap induced by permanent magnets can be
approximated using the Fourier series [13,15,29–32]:
approximated using the Fourier series [13,15,29–32]:
where the Fourier spectrum contains harmonics of ςth order, which belong to the set
Q  ...  5p, 3p,  p, p, 3p, 5p,... .

X
jς(θ−ϕ)
Bm (Bθm −
( ϕ,
 r,)r )= BBPMPM
e j (    )
ς ((rr) ) ·e (1) (1)
ς∈Q
 Q

wherewhere the Fourier spectrum contains harmonics

of of
ςthς order,
order,which
whichbelong
belong totothe
theset
setQ =
th
the Fourier spectrum contains harmonics
Q  ...  5p, 3p,  p, p, 3p, 5p,...
.
. . . − 5p, −3p, −p, p, 3p, 5p, . . . .


Fourier coefficients BPM ς (r) can be defined using a two-dimensional (2D) model of the magnetic
field distribution according to [13,15]. The 1D models in this case are too imprecise and do not provide
sufficient quantitative correspondence.

2.1.2. Distribution of Flux-Density in a Machine with Real Shapes of a Magnetic Circuit

An illustration of the methodology of deriving an analytical model of the magnetic flux density
distribution in the air gap was made for the model with cores [32], presented in the coordinate
system according to Figure 6. Basing on this model, any other shapes of the magnetic circuit can also
be inserted.
 2.1.2. Distribution of Flux-Density in a Machine with Real Shapes of a Magnetic Circuit
2.1.2. Distribution of Flux-Density in a Machine with Real Shapes of a Magnetic Circuit
An illustration of the methodology of deriving an analytical model of the magnetic flux density
An illustration
distribution in the airofgap
the was
methodology of deriving
made for the an analytical
model with cores [32],model of the
presented in magnetic flux density
the coordinate system
distribution in the air gap was made for the model with cores [32], presented in the coordinate
according to Figure 6. Basing on this model, any other shapes of the magnetic circuit can also system
be
according
inserted. to Figure 6. Basing on this model, any other shapes of the magnetic circuit can also be
Energies 2020, 13, 5741 5 of 28
inserted.

rotor 1 am
rotor 1 ac
am 
ac 
S PM N S N lm
S PM N z S   N lm
z  
1 2' 2 3' 3 1' 0z 1 3'
2' 32 3' l i l
1 2' 2 3' 3 1' 0 z 1 2'
3' 32 3' l i l
0
a sc 0
m N
a sc S N S lm
m N S N S lm
rotor 2 
rotor 2 

Figure Figure 6. Model

6. Model for the
for the analysis
analysis ofofAxial
AxialFlux
Flux Permanent
PermanentMagnet
MagnetGenerator (AFPMG)
Generator magnetic
(AFPMG) magnetic
Figure 6. Model for the analysis of Axialdistribution
flux-density Flux Permanent Magnet
with stator Generator (AFPMG) magnetic
cores.
flux-density distribution with stator cores.
flux-density distribution with stator cores.
Figure 7 shows an example of windings arrangement [8] and placement of the iron cores inside
Figure 7 shows an example of windings arrangement [8] and placement of the iron cores inside
the coils. 7 shows an example of windings arrangement [8] and placement of the iron cores inside
Figure
the coils.
the coils.

1
(r )
1
(r )
3 2 r
3 2 r

lc Ri Ro
lc Ri Ro

(a) (b)
(a) (b)
Figure
Figure 7. (a)7.Assumed
(a) Assumed winding
winding arrangement; (b)
arrangement; (b) the
the distribution
distributionofof
the iron
the cores
iron inside
cores the coils.
inside the coils.
Figure 7. (a) Assumed winding arrangement; (b) the distribution of the iron cores inside the coils.
The model
The model of magnetic
of magnetic flux-density
flux-density distribution
distribution in the in air-gap
the air-gap
of anofAFPM
an AFPM generator
generator is a is a
function
The of
function model
four of magnetic
variables flux-density
(dependent on a distribution
axial component in the
z, air-gapaccording
location of an AFPMto thegenerator
stator θ, is a
angle
of four variables (dependent on a axial component z, location according to the stator θ, angle of the
function of four variables (dependent on a axial component z, location according to the stator θ, angle
rotor of ϕ and
the rotor position φ and radialRlocation R  r  R ). It was assumed that the general equation
position radial location
of the rotor position φ and radial location i ≤ r ≤ RR o ).
i It waso assumed that the general equation describing
 r  R o ). It was assumed that the general equation
i
describing the
the distribution of distribution of magnetic flux-density
magnetic flux-density is simplifiedisand simplified and is considered
is considered onlyaxial
only for the for the axial
component
describing
component the
z = 0. It means that: z 
distribution
0 . It means ofthat:
magnetic flux-density is simplified and is considered only for the axial
component z  0 . It means that:
Bδ (Bz, (θ,
z, ϕ,, r), r=) BBm ((z,z ,θ,
 ,ϕ,
 , rr)) =  ,θ,
 BB((zz=0,0,  ,ϕ,
r ) r) (2) (2)
Bδ ( z ,  ,  , r )  Bm ( z ,  ,  , r )  B( z  0,  ,  , r )
δ m
(2)
By applying
By applyingAmpere’s
Ampere’scircuital law law
circuital for the
for contour
the contour marked in Figure
marked 6 and
in Figure 6 the
andassuming
the assuminglinearity
By
of thelinearityapplying
magnetic Ampere’s
circuit,
of the one circuit,
magnetic circuital
can getone thecan law for
equations the contour
describing
get the equations marked
magnetic
describing in Figure 6 and the
field distribution,
magnetic assuming
wellwell
field distribution, known
from linearity
classical
known ofconsiderations:
from the magnetic
classical circuit, one can get the equations describing magnetic field distribution, well
considerations:
known from classical considerations:
B(z = 0, θ, ϕ, r) = B(θ, ϕ, r) = BΘ (θ, r) + BPM (θ − ϕ, r) (3)

where:
BΘ (θ, r)—axial component of the flux-density distribution resulting from winding currents (windings
magnetomotive forces MMF),
BPM (θ − ϕ, r)—axial component of flux-density from permanent magnets.
l (θ−ϕ,r)
By assuming l0m (θ − ϕ, r) = m µ and defining the unit permeance function as a magnitude
rm
proportional to the inverse of the magnetic field force lines length:
µ0
λ(θ, ϕ, r) = (4)
lδ (θ, r) + l0m (θ − ϕ, r)

Function lδ (θ, r) in case of the cores placed inside the stator coils, can be approximated as shown
in Figure 8.
 magnitude proportional
magnitude to the
proportional inverse
to the of the
inverse magnetic
of the field
magnetic force
field lines
force length:
lines length:
μ 0 μ
 (, 
(,,r) ,r )  0 (4) (4)
lδ (l , (r) ,r l)m' (l ' (, r) , r )
δ m

l ( , r ) in case of the cores placed inside the stator coils, can be approximated as
13, 5741δ lδ ( , r ) in case of the cores placed inside the stator coils, can be approximated as6 of 28
Function
Function
Energies 2020,
shown in Figure
shown 8. 8.
in Figure

    3 3 2 2
0 0 2ps 2p ps p 2p s 2p
s s s ps p
s

Figure 8. Equivalent function of the magnetic field force lines length in the air gap.
Figure 8. Equivalent functionofofthe
themagnetic
magnetic field
Figure 8. Equivalent function fieldforce
forcelines
lineslength in the
length air gap.
in the air gap.

And thethe
function l (   , r ) , rcan be approximated as shown in Figure 9:
And And
the function lmm(θlm−(ϕ, 
function r) can) can
be be approximated
approximated asasshown
shown in
in Figure
Figure9:9:

l
lm(θ (θ
m
, r), r)

lm l
m

l m /l rm
m /  rm
θ 
θ 
-r) r)r)
/2p /p/p
-/p-/p -/2p -r)
-/2p /2p

Figure9.9.Equivalent
Figure functionofofthe
Equivalent function the magnetic
magnetic field
field force
force lineslines length
length in permanent
in permanent magnets.
magnets.
Figure 9. Equivalent function of the magnetic field force lines length in permanent magnets.

ByBy
defining
definingthe
thewinding magnetomotive
winding magnetomotive force
force (MMF)
(MMF) function
function as: as:
By defining the winding magnetomotive force (MMF) function as:
 2
θZ+2π  2
s  
 , r ,r  as a , r d, r d
Θs (θ, r) =
   , r d, r d0  0
Θs (θ, r)dθ ≡ 0
Z
(5) (5)
as (θ, r)dθ
s
s s

s
(5)


where a s (a, r() —stator ampere-turns, thethe

axial θ of flux-density can be defined as follows:
components
where s , r ) —stator ampere-turns, axial components of flux-density can be defined as follows:
where as (θ, r)—stator ampere-turns, the axial components of flux-density can be defined as follows:

θ+
R 2π
λ(θ0 , r)·Θs (θ0 , r) dθ0
θ
BΘ (θ, r) = λ(θ, r) {Θs (θ, r) − } (6)
θ+
R 2π
λ(θ0 , r) dθ0
θ

θ+
R 2π
λ(θ0 , ϕ, r) ·Bm (θ0 − ϕ, r) dθ0
λ(θ, ϕ, r) θ
BPM (θ, ϕ, r) = { Bm (θ − ϕ, r) − } (7)
λref (r) θ+
R 2π
λ(θ0 , ϕ, r)dθ0
θ

where:

λref (r) = λ(0, 0, r) is the reference magnitude for the origin of the coordinate system,
Bm (θ − ϕ, r) is the flux-density distribution induced by permanent magnets for a coreless machine.

The obtained Equations (6) and (7) are modifications of the well-known classical equation
describing the one-dimensional distribution of the magnetic field in the air gap [33,34]. The presence
of permanent magnets in the machine magnetic circuit causes some modifications in the form of an
additional component BPM (θ, ϕ, r) [29,35] described by the Equation (7).
 Bm (   , r ) is the flux-density distribution induced by permanent magnets for a coreless machine.

The obtained Equations (6) and (7) are modifications of the well-known classical equation
describing the one-dimensional distribution of the magnetic field in the air gap [33,34]. The presence
of permanent magnets in the machine magnetic circuit causes some modifications in the form of an
additional
Energies component BPM ( , , r ) [29,35] described by the Equation (7).
2020, 13, 5741 7 of 28
The magnetomotive force (MMF) of a 3-phase symmetrical AFPM winding (the arrangement of
the stator windings is shown in Figure 7) can be described by the following relationship [29,34,35].
The magnetomotive force (MMF) of a 3-phase symmetrical AFPM winding (the arrangement of
3

 ,be
the stator windings is shown in Figure 7)can r  described
   , by
s 
r  the following relationship [29,34,35].
a 1
a (8)
3
The Fourier distribution of the magnetomotive (Figure 10) force for the winding “a” is presented
X
Θs (θ, r) = Θa (θ, r) (8)
by the following equation:
a=1

a  , r    (Figure
v  
jv   a 
The Fourier distribution of the magnetomotive r  e 10) force for the winding “a” is presented
a
(9)
vP
by the following equation:
where the Fourier spectrum of the magnetomotiveX force MMF contains harmonics of vth order,
Θa (θ, r) = Θav (r)·e jv(θ−θa ) (9)
belonging to the set P  ...  3p s , 2p s , p s , p s , v∈P
2p s , 3p s ,... .
2π
where theFor
Fourier
3-phasespectrum
winding,  a magnetomotive
ofa the 1  1, 2,3
, where a force oMMF
. contains harmonics of vth order,
3p
n
belonging to the set P = . . . − 3ps , −2ps , −pss , ps , 2ps , 3ps , . . . .

Figure 10. Winding “a” magnetomotive force (MMF) distribution.

Figure 10. Winding “a” magnetomotive force (MMF) distribution.

For 3-phase winding, θa = (a − 1) 3p 2π

, where a = 1, 2, 3.
The coefficients in Equation (9) ares as follows:
The coefficients in Equation (9) are as follows:

1
Θav (r) = ia Wvs (r) (10)
π

w ·k|v| (r)
where: Wvs (r) = s |v|s , ks|v| —the winding factor for vth harmonic and ws —the total number of turns
of the stator winding. For concentrated coils, the winding factor [8,29,34,35] can be defined as:

sc α (r)
ε(r) sin(|ν| 2 )
ks|ν| (r) = sin(|ν| )· (11)
2 α (r)
|ν| sc 2

aco +aci
ε(r) = arc —an angle of coil pitch or coil span at coordinate r, ac ≈ 2 .
αsc (r) = arsc —an angle of the coil side width at coordinate r.

The above equations are valid for both coreless and core stators. However, in case of coreless
machines, the components of the flux-density distribution have a simpler form.

2.1.3. Flux-Density Distribution in Stator Coreless Machine

The unit permeance, in general, is a function dependent on the coordinates θ, ϕ and r. For the
machine model with a coreless stator (Figure 11), the equations described in the previous sections are
significantly simplified.
 machines, the components of the flux-density distribution have a simpler form.

2.1.3. Flux-Density Distribution in Stator Coreless Machine

The unit permeance, in general, is a function dependent on the coordinates θ, φ and r. For the
machine
Energies 2020, 13,model
5741 with a coreless stator (Figure 11), the equations described in the previous sections are8 of 28
significantly simplified.

rotor 1 am
ac 
S PM N S N lm
z 

1 2' 2 3' 3 1' 0 z 1 3'

2' 32 3' l
0
a sc
m N S N S lm

rotor 2 

Figure 11.11.Cross
Figure Crosssection
section of corelessAFPMG
of coreless AFPMGconstruction.
construction.

Assuming again
Assuming thatthat
again thethe
relative
relativemagnetic
magneticpermeability
permeabilityofof permanent magnetsisissimilar
permanent magnets similartotothe
the air
magnetic permeability (µ
air magnetic permeability  . . .
rm ( μ rm  1.01...1.1 ) for the case of a coreless machine, the distribution of the unit
1.01 1.1) for the case of a coreless machine, the distribution of the
permeability function can be approximated only by the constant component:
unit permeability function can be approximated only by the constant component:

λ0 = µ00 (12)

0 +2l2l' 0m
llδ  (12)
δ m

lm
where: l0m
= .
'µrm l m
where:
The MMF
l m component
. describing the field distribution, depending on the winding currents for a
 rm
3-phase AFPM machine, can be presented as [29–32]:
The MMF component describing the field distribution, depending on the winding currents for a
3-phase AFPM machine, can be presented as [29–32]:
3 XX
BΘ (θ, r) = λ03 · Θa (jr()·e) jν(θ−θa ) (13)
B  , r    0  
a=1 ν∈P
a 1 P
ν
 (r )  e a

a
(13)

The form of a component from permanent magnets is also simplified:

X
jς(θ−ϕ)
BPM (θ, ϕ, r) = Bm (θ − ϕ, r) = BPM
ς (r)·e (14)
ς∈Q

The coefficients of a 2D magnetic flux-density component distribution from permanent magnets

(the distribution preferred for the coreless stator) in the middle of the air gap according to [13,15,29–32]
are as follows:
l 2 l +l
2 Br p 2sinh(ς mr )· cosh(ς m2r δ )
BPM
ς ( r ) = · sin ( ς·β ( r )) (15)
π ς µ ·sinh(ς δ
l +2 lm
) rm r
am amo +ami
where: β(r) = 2r ; am ≈ 2 .

2.1.4. Flux-Density Distribution in Machine with Cores

In relation to a coreless machine, the unit permeability for a machine with iron cores is a function
of the radial coordinate r and a coordinate θ associated with the stator. If we re-assume that the relative
magnetic permeability of PM is close to the magnetic permeability of air (µrm  1.01 . . . 1.1) the unit
permeance function is independent of the rotor position (reference system to the first winding first
coil axis).
The Fourier distribution of the unit permeance function (Figure 12) can be therefore defined
as (16): X
λ(θ, r) = λm (r) ·ejmθ (16)
m∈M
  ( , r )   m (r )  e jm (16)
mM

where:

 3  (r )   sc (r )
Energies 2020, 13, 5741  λ min  (λ max  λ min ) π ps 2
for m  0
9 of 28

m (r )   (17)
 3 ps  (r )   sc (r )
where: (λ max  λ min ) sin(m ) for m  M  {0}
  λ min + ( π
λ m
max − λ ) 3 2 ε(r)−αsc (r)
min π s p for m = 0
2


λm ( r ) = 

μ0 μ0 (17)

 λ max  ελ )−αsc (r)

 ps ' (rmin (18)
 (λmax − λlδmin li) π
3 2l
mm sin ( m l 
2 δ ) 2l for
m
m ∈ M − {0}

m-order of the λ permeance µ0 harmonic µ0belonging to the set (18)

max = 0 λmin =
M  ...  9p s , 6p s , 3p s , 0, 3p s , 6p s , 9p s ,...
lδ−. li + 2lm lδ + 2lm

 ( , r )
(r) sc(r)
 max

 min

  3 2
0 2ps ps 2p s ps
1 2 3 1

Figure 12. Unit permeance function AFPMG with stator cores.

Figure 12. Unit permeance function AFPMG with stator cores.

n m-order of the permeance o harmonic belonging to the set M =

. . . − 9ps , −6ps , −3ps , 0, 3ps , 6ps , 9ps , . . . .
Magnetic flux density distribution induced by PM for AFPMG with iron cores can be analyzed
using the unit permeance function and the base flux density distribution for a coreless machine.
The equation describing the distribution of the flux-density component depending on the winding
currents can be simplified according to the Equation (6) by eliminating the constant component
according to the following equation:

λm (r) Θaν (r) for m + ν = 0

 P P
θZ
+2π 
 2π θZ
+2π
m∈M ν∈P


λ(θ0, r)·Θa (θ0, r) dθ0 =  λ(θ0, r)dθ0 = 2π λ0

(19)




θ 0for m + ν , 0 θ



Finally, one can obtain the Equation (20) describing the flux density component, dependent on the
winding currents, in the form:

3 X X
X
BΘ (θ, r) = c(m, ν)·λm (r)·Θaν (r) · ej(m+ν)θ · ej(−νθa ) (20)
a=1 m∈M ν∈P

where:
0 for m + ν = 0



c(m, ν) = 

(21)


 1 for m + ν , 0


Similar simplifications can be assumed for the flux-density component excited by PM (7). Finally,
it can write:
X X λ (r) PM
BPM (θ, ϕ, r) = c(m, ς) m B (r)·ej(ς+m)θ ·ej(−ς) ϕ (22)
λref (r) ς
m∈M ς∈Q

The Fourier spectrum coefficients of the flux-density distribution excited by PM, appearing in
Equation (22) can be presented for a two-dimensional (2D) field distribution model in the
form [13,15,29–32]:
l 2l +l −li
2 Br p 2 sinh(ς mr )· cosh(ς m 2rδ )
BPM
ς (r) = sin(ς·β(r)) (23)
π ς 2 l +l −li
µ ·sinh(ς m δ )rm r
Energies 2020, 13, 5741 10 of 28

2.2. Generator Model Equations

The usage of Lagrange’s formalism, based on the characteristics of the windings and the elements
co-energy function of the entire electromechanical system, allows for a mathematical modelling of the
AFPMG. Basing on [29–35] the equations of a permanent magnet machine with a three-phase winding,
(assuming sign convention for passive component) can be defined in a standard matrix form:

 ψPM1 (ϕ)
         
 Lσs + Lss   i1   i1    u1 
 d 
  i2  + Rs ·  i2  + d  ψPM2 (ϕ)
       
 Lσs + Lss  dt 
 =  u2  (24)
    dt    
Lσs + Lss i3 i3 ψPM3 (ϕ) u3
        

d2 ϕ dϕ
2
= TL + Tem (i1 , i2 , i3 , ϕ) + Tcog (ϕ) − D
J (25)
dt dt
where the electromagnetic torque Tem and cogging torque Tcog :

i ∂  ψPM1 (ϕ)

 
h 
Tem (i1 , i2 , i3 , ϕ) = i3 ·  ψPM2 (ϕ)

i1 i2  (26)
∂ϕ  
ψPM3 (ϕ)


∂E0PM (ϕ)
Tcog = (27)
∂ϕ
E0PM (ϕ)—co-energy component independent of winding currents.
The above equations are quite obvious, and their structure is very similar to the Lagrange equations
for conventional electrical machines. Due to the presence of permanent magnets in the machine
magnetic circuit, some adjustments should be made while determining the basic parameters in relation
to classic models.
In order to describe the basic parameters of the generator mathematical model (24), the winding
flux linkage was defined. The equation describing the flux ψa (ϕ), linked with the winding “a” can be
defined as:  ε(r)−αsc (r) 
ZRo 
 2Z + θa 


 

ψa ( ϕ ) = θ, ϕ,
 
ws  B ( r ) dθ r dr (28)
 
 


 

Ri −ε(r)+αsc (r)
 
+ θa

2

R +R
In order to simplify, some averaging has been made by assuming that r ≈ rs = o 2 i . Then flux
ψPMa , linked with winding “a”, generated by permanent magnets in zero current state, is described by
standard relation:
jς{(a−1) 4π
X
ψPMa (ϕ) = ψPMs ·e 3p −ϕ} for a = 1, 2, 3 (29)
ς
ς∈Q

while the generator electromotive force (EMF) for winding “a” (in zero current state) at constant speed,
is determined by the relationship:
X 4π
ePMa = Eς · cos ς{Ωt − (a − 1) }for a = 1, 2, 3 (30)
3p
ς=p,3p,5p...

where: Eς = 2ς Ω ψPMs ς .
The leakage inductances are analytically expressed as the sum of two components [8]. The first
one is related to the leakage flux around the active conductors’ radial part (coil sides) and the second
Energies 2020, 13, 5741 11 of 28

one depends on the leakage flux around the end windings (top and bottom of the coil). The leakage
inductance coefficient can be determined from following equation:

Lσs ≈ 2µ0 · (ws )2 [l0c + (ac − asc )] ·0.3/ps (31)

where l0c is the length of the active side of the coil and in most cases l0c ≈ Ro − Ri .

2.2.1. Model of Coreless Generator with Simple Magnets

The distribution function coefficients of the flux from permanent magnets linked with the stator
winding (29) must be modified due to the weakening of the flux at the magnets’ edges. This correction
is important for the quantitative analysis due to the relatively large air gap in the analyzed structure of
discussed generator with a coreless stator.
Considering a two-dimensional distribution of flux-density in the air gap (taking into account
Energies 2020, 13, x FOR PEER REVIEW 12 of 29
the Equation (15)) it is possible to model the effects on the magnets side edges along the coordinate
θ. Aareseparate problem
perpendicular is to
to the include rthe
coordinate weakening
(Figure 13), that of
is, the permanent
on the inside (formagnets fluxouter
r  R i ) and at the
(foredges
that are perpendicular to the coordinate r (Figure 13), that is, on the inside (for r = Ri ) and outer
r  R o ) magnets edges. The proposed analytical equations allowing us to derive appropriate
(for r = Ro ) magnets edges. The proposed analytical equations allowing us to derive appropriate
correction functions are not quite as complex [13,15]. In this paper, a simpler approach is proposed.
correction functions are not quite as complex [13,15]. In this paper, a simpler approach is proposed.
Approximate distribution of the magnetic flux-density produced by PM in the middle of the air gap
Approximate distribution
( z  0 ), along the radial of the magnetic
coordinate, flux-density
is shown produced by PM in the middle of the air gap
in Figure 13.
(z = 0), along the radial coordinate, is shown in Figure 13.

Bm ( 0, r) lm  lδ
2
B max

Bedge lc
r
0
Ri Ro
Figure 13. Approximate distribution of the magnetic flux-density produced by permanent magnets in
Figure 13.
the middle Approximate
of the air gap (z distribution
= 0) along of the magnetic
coordinate “r” flux-density
for θ − ϕ = produced
0. by permanent magnets
in the middle of the air gap ( z  0 ) along coordinate “r” for    = 0 .
The authors proposed to introduction a correction factor ke , which represents the flux weakening
Theand
at the inner authors
outerproposed
magnets to introduction
edges. a correction
This coefficient factor
can be k e , which
determined represents
as the ratio: the flux
weakening at the inner and outer magnets edges. This coefficient can be determined as the ratio:
Br (AV)
ke = Br (AV) (32)
ke  B (32)
Bθ( AV)
(AV)

where:
where:

Br (AVB) —the
r ( AV) —the average
average value
value of the
of the axial
axial componentofofthe
component themagnetic
magnetic flux-density
flux-densitydistribution
distributionfrom PMPM in
from
in the middle
the middle of the
of the air gapair ((θ −
Bmgap (() =
Bmϕ  0,
) r)0,along
r ) along
thethe coordinate
coordinate r (Figure
r (Figure 13)
13) ininthe
therange
rangeofof(R( , Ro ),
i
Bθ (AVR)i—the
, R o ), average value of the axial component of the magnetic flux-density distribution from PM
in theBmiddle of the
 (AV) —the air gap
average valueBmof(θthe ϕ, rs )component
− axial (Equationof(1),
theFigure 5) according
magnetic flux-densitytodistribution
the coordinates θ − ϕ in
from PM
the range
in the middle of
, β(air
of (−β(rs )the rs ))gap
, B (   , r ) (Equation (1), Figure 5) according to the coordinates   
m s
Bmax —maximum flux-density value,
in the range of (   (rs ),  (rs )) ,
Bedge —the value of the magnetic flux-density at the edge of the magnets.
Bmax —maximum flux-density value,
Bedge —the value of the magnetic flux-density at the edge of the magnets.
Bedge , Bmax values (Figure 13) can be determined both by finite element analysis and by using
analytical equations. Using the analytical waveform approximation (Figure 13) for the cases where
l c  (l m  l δ ) one can get:

lc - (l m  lδ ) l m  lδ 2 l m  lδ
Br (AV)  Bmax  Bedge  ( Bmax  Bedge ) (33)
lc lc π lc
Energies 2020, 13, 5741 12 of 28

Bedge , Bmax values (Figure 13) can be determined both by finite element analysis and by using
analytical equations. Using the analytical waveform approximation (Figure 13) for the cases where
lc > (l m + lδ ) one can get:

lc − (l m + lδ ) lm + lδ 2 lm + lδ
Br (AV) = Bmax + Bedge + (Bmax − Bedge ) (33)
lc lc π lc

where: X
Bmax = BPM
ς (rs ) (34)
ς∈Q

lc = Ro − Ri is the magnet length.

In case of analytical considerations, assuming that lc > (l m + lδ ) and considering:

1
Bedge = B (35)
2 max
the Equation (33) is reduced to a very simple form:

2 lm + lδ
Br (AV) = Bmax ·[ 1 − (1 − ) ] (36)
π 2 lc

Basing on the analytical equations of magnetic flux-density distribution induced by PM in the

middle of the air gap (1), it is possible to determine the average value of this function in the range
(−β(rs ), β(rs )):
X sin ς β(r s )
Bθ (AV) = BPM
ς ( rs ) · (37)
ς β(r s )
ς∈Q

The corrected form of coefficients of flux linkages induced by PM (29) is presented by the
following equation:
ψPMs
ς = 2 ke ·BPM s
ς (rs ) ·Wς (rs ) ·rs ·lc (38)

After formal mathematical operations, the windings inductances can be described using standard
relationships [8,34,35]:
X 2
Lss = Lss ss
ν Lν = ·[Wνs (rs )]2 ·rs ·l0c ·λ0 (39)
π
ν∈P

Co-energy function for a coreless machine in a zero current state is constant (E0PM (ϕ) = const)
and therefore the cogging torque does not occur (Tcog (ϕ) = 0).

2.2.2. Model of a Generator with Stator Cores and Simple Magnets

While deriving the flux linkage distribution coefficients (29), for the generator model with cores,
it can be assumed that the problem of the magnet flux weakening at the inner and outer magnets
edges is not so significant and there is no need to take these effects into account (ke = 1). After formal
mathematical transformations, the coefficients of the flux linkage function distribution take the
following form:
X λ (rs ) PM
ψPMs
ς = 2 c(m, ς) m ·B (rs ) ·Wςs +m (rs ) ·rs ·lc (40)
λref (rs ) ς
m∈M

The windings inductances can be defined as follows [29–32,34,35]:

X X2
Lss = c(m, ν) ·Wνs (rs ) ·Wνs+m (rs ) ·rs ·l0c ·λm (rs ) (41)
π
m∈M ν∈P
Energies 2020, 13, 5741 13 of 28

For the machine with cores, the co-energy component, independent of the winding currents,
should be analyzed. This component is related to the energy stored in the air gap and in permanent
magnets [35,36]. The following analysis is based on the magnetic flux-density distribution equations in
the zero-current state for the base model and the unit permeance function. The co-energy component
in zero current state can be defined as [35,36]:

ZRo Z2π Z2π

[BPM (θ, ϕ, r)]2 lc rs
E0PM (ϕ) = { dθ} r dr ≈ [λ(θ, rs ) Bm (θ − ϕ, rs )2 ] dθ (42)
λ(θ, r) 2 λref (rs )2
Ri 0 0

The function of squared magnetic flux density from PM, occurring in Equation (42), can be
approximated by the function:
X
Bm (θ − ϕ)2 = BBm k ·ejk(θ−ϕ) (43)
k∈K

The Fourier distribution coefficients of this function belong to the set K =

. . . − 6p, −4p, −2p, 0, 2p, 4p, 6p, . . . . The Fourier decomposition coefficients of the squared magnetic


flux density function are obtained using analytical Equations (1), (14) and the FFT procedure.
Determining the integral in (42), the following relationship can be obtained:

Z2π 
 2π
P P
λm (rs ) ·BBm k · e j(−k)ϕ for k + m = 0
2
[λ(θ, rs ) Bm (θ − ϕ, rs ) ] dθ = 

k∈K m∈M (44)

0 for k + m , 0


0

After performing the formal mathematical transformations, a general equation defining co-energy
in a current less state for AFPMG with stator cores is obtained:
π ·lc · rs X
E0 PM (ϕ) = Real { λ−q (rs )· BBm q ·e−jqϕ } (45)
λref (rs )2 q∈K∩M

An important phenomenon in case of machines with PM is the presence of cogging

torque [25–28,35,36]. For a model with cores, these torques will occur. The cogging torque can
be described as:
π·lc ·rs X
Tcog (ϕ) = − Imag { (−q)·λ−q (rs )·BBm q ·e−jqϕ } (46)
λref (rs )2 q∈K∩M

2.2.3. Model of a Generator with Skewed Magnets

For the cases of skewed magnets arrangement, the co-energy function and associated relations
representing the windings linkage fluxes require some adjustments. In classical machines, adjustments
resulting from the skew of slots on the stator or rotor are made using the so-called skew factor. This can
also be taken into account for disc machines with PM.
Assuming of permanent magnets skew (according to Figure 14), it is necessary to correct the
origin of the reference system related to the rotor (coordinate ϕ) relative to the radial coordinate r with
a reference selection of the middle magnet length.

r − rs
ϕ(r) = ϕ + αsk (47)
lc
ask Ro + Ri
where αsk —skew angle of permanent magnets; αsk ≈ q , lc = Ro − Ri ; rs = 2 .
R2o +a2sk
 r  rs
 (r )    α sk (47)
lc

a sk Ro  Ri
where α sk —skew angle of permanent magnets;  sk  , lc  R o  R i ; rs  .
Energies 2020, 13, 5741 R a 2
o
2
sk
2 14 of 28

a sk
Ro

 (r ) lc / 2
PM r
 rs

lc / 2

Ri
0θ

(a) (b)
Figure
Figure 14. Cross-section
14. Cross-section ofofa askewed
skewedPM:
PM: (a)
(a) reference
referenceframe;
frame;(b)(b)
rotor disc
rotor with
disc skewed
with PM. PM.
skewed

Based on the Equations (14) and (22) it is possible to modify the general dependencies determining
the distribution of magnetic flux-density from PM in the AFPM machine with a magnets skew,
which can be written as:
X X λm (r) PM
BPM (θ, ϕ(r), r) = c(m, ς) B (r)·ej(ς+m)θ ·ej(−ς) ϕ(r) (48)
λref (r) ς
m∈M ς∈Q

0 for m + ς = 0
(
where: c(m, ς) =
1 for m + ς , 0
Flux ψPMa from PM linked with winding “a” in zero current state, should be refactored into
below equation:
 ε(r) 
 2 + θa
ZRo 



 Z 

ψPMa (ϕ) = θ, ϕ
 
ws  B ( ( r ) , r ) dθ r dr (49)
 
 PM 


 

Ri −ε(r)
 
+ θa

2

In order to simplify the integration (Equation (49)) the authors assumed an average value r ≈ rs
λm ( r )
for λref (r)
; BPM
ς (r). Integration over the coordinate “r” takes the form of the following expression:

ZRo αsk
j ς r−rs sin (ς 2 ) |ς|
e lc αsk dr = lc = lc ksk (50)
αsk
ς 2
Ri

|ς|
from which the skew factor of the permanent magnets ksk is obtained [36].
For winding “a” the flux linkage ψPMa from PM in zero current state, can be defined as:
X
ΨPMa (ϕ) = ·ΨPMs
ζ ·ejζ(θa −ϕ) fora = 1, 2, 3 (51)
ζ∈Q

where:
X λm (rs ) PM sp |ζ|
ΨPMs
ζ = 2c(m, ζ) ·Bζ (rs )·Wζ+m (rs )·ksk ·rs ·1c (52)
m∈M
λref (rs )
|ς| αsk
sp ws ·ksp (r) |ς| sin (ς 2 )
Wς ( r ) = ; ksk = αsk (53)
|ς| ς 2
Energies 2020, 13, 5741 15 of 28

Similar adjustments should also be made to the co-energy component (42) independent of the
winding currents:
ZRo Z2π h i
E0PM (ϕ) = { λ(θ, r) Bm (θ − ϕ(r), r)2 dθ} r dr (54)
Ri 0

After integrating Equation (54) and taking into account the Equation (48), the following relation
can be obtained:

Z2π  P P
λm (r)·BBmk ·e j(−k)ϕ(r) for k + m = 0
 2π

2
[λ(θ, r) Bm (θ − ϕ(r), r) ] dθ = 

k∈K m∈M (55)
0 for k + m , 0


0

Considering the defined forms of the unit permeance function (4), the magnetic flux density square
(43) for the simplified case when r ≈ rs and performing formal mathematical operations, the equation
describing the co-energy for the zero current state in a modified form can be obtained:

π·lc ·rs X
|q|
E0PM (ϕ) = 2
Real { λ−q (rs )·BBmq ·ksk ·e−jqϕ } (56)
(λδm ) q∈K∩M

which contains the skew factor of the magnets.

From the Equation (56), one can get the modified form of Equation (46) which describes the
cogging torque of a generator with magnets placed obliquely:

π·lc ·rs X
|q|
Tcog (ϕ) = − 2
Imag { (−q)·λ−q (rs )·BBmq ·ksk ·e−jqϕ } (57)
(λδm ) q∈K∩M

By using the appropriate magnets skew angle, one can work on effective reduction of the cogging
torque. By analyzing the relationships presented above, it can be seen that the set of squared magnetic
flux density function harmonics for a coreless machine K = . . . − 6p, − 4p, − 2p, 0 , 2p , 4p , 6p . . . ,


while the set of unit permeance function harmonics M = {. . . − 3Zs , − 2Zs , − Zs , 0, Zs , 2Zs , 3Zs , . . .}
and, therefore, the smallest common part of the sets K and M different from 0 is equal 6p = 4Zs , so the
fundamental harmonic of the cogging torque for the analyzed generators will always correspond to
the number 6p (fourth permeance harmonic).

3. Simplified, Monoharmonic AFPMG Model in Steady State

The steady-state generator model assumes the presence of all significant spatial harmonics.
This is reflected in the sets of winding currents harmonics, and thus also in the electromagnetic
torque [29]. The purpose of the considerations in this chapter is to refer to the classic models of electric
machines and show the structure of the simplified model, which is valid assuming the presence of
only basic harmonics of the magnetic field distribution and the internal and external symmetry of the
machine operating in the steady-state, which means that the rotor angular speed is constant (ω = Ω).
Therefore, the generator model can be reduced to the following equivalent diagram (Figure 15) in
which there are phase quantities (RMS) of the voltage (EMF) of the grid and the generator:
 [29]. The purpose of the considerations in this chapter is to refer to the classic models of electric
machines and show the structure of the simplified model, which is valid assuming the presence of
only basic harmonics of the magnetic field distribution and the internal and external symmetry of the
machine operating in the steady-state, which means that the rotor angular speed is constant ( ω = 
). Therefore, the generator model can be reduced to the following equivalent diagram (Figure 15) in
Energies 2020, 13, 5741 16 of 28
which there are phase quantities (RMS) of the voltage (EMF) of the grid and the generator:

PM GENERATOR LOAD

XS Rs IG RL XL

EGph  ej(0) UG ES ph  e j 0

Figure
Figure 15.15. Simplifiedequivalent
Simplified equivalent phase
phasegenerator
generatordiagram.
diagram.

By defining: √
EG ph = pΩ 2 ψPM p
s; R
SL
= Rs + RL ; XSL = XS + XL ; XL = pΩ LL ; Xs = pΩ (Lσs + Lss ).
The equation for phase current (RMS) calculation is as follows:

ES ph ·ej β0 − EG ph ·ej(ϑ+β0 )
IG = (58)
RSL + j XSL

where ϑ determines the power angle, while the equation describing the electromagnetic torque is
as follows:
3E2G ph ES ph
Tem(AV) = − q {cos(α0 ) − cos(ϑ + α0 )} (59)
EG ph
Ω R2SL + XSL 2

X
where: α0 = arctan( RSL )
SL
Equation (59) can also be defined as follows:

3E2G ph ES ph
Tem(AV) = − [RSL − {RSL cos(ϑ) − XSL sin(ϑ)}] (60)
Ω(R2SL + XSL
2 ) EG ph

The above dependencies may be useful to determine the generator’s operating parameters in
the steady state as well as to determine the unknown power angle value. The simplified equation
describing the value of the machine electrical power can be presented in the form:

∨  2
Pel = 3 Real {IG ·ES ph ·ej β0 − RL IG  } (61)

In case of a standalone generator state operation, the generator phase current, electromagnetic
torque and processed electrical power equations can be described by very simple forms:

−EG ph ·ejϑ
IG = (62)
RSL + j XSL

3E2G ph RSL
Tem(AV) = − (63)
Ω(R2SL + XSL2 )

 2
Pel = −3 RL IG  (64)
Energies 2020, 13, 5741 17 of 28

According the above equations one can see that for analyzes of standalone generator operation
state with RL load, there is no need to seek power angle. The power angle which is defined as the
angle between the generator EMF and the load voltage will be equal to:

XL
ϑ = α0 − α0L where α0L = arctan( ) (65)
RL

The application of a simplified model also allows for the analysis of generator operation with
a 6-pulse bridge diode rectifier load [37–39]. Due to the relatively low generator reactance, one can
assume a rectifier model, in which 3 diodes conduct during the commutation period, while only two
diodesEnergies
operate during the conduction period.
2020, 13, x FOR PEER REVIEW 18 of 29
The commutation angle is then τ ≤ π3 which corresponds to condition fulfilment:

 I 3E 3EG ph
G ph (66)
IG  G≤ 2π X (66)
2π XSS
In that case, a simplified equivalent circuit is valid (Figure 16).
In that case, a simplified equivalent circuit is valid (Figure 16).

PM GENERATOR
WITH DIODE RECTIFIER LOAD

RDC
IG I DC

IG  ej
UDC RL
EG ph EDC

Figure 16. Simplified

Figure equivalent
16. Simplified equivalentdiagram
diagramof
of aa generator loadedwith
generator loaded with a diode
a diode rectifier.
rectifier.

The parameters used

The parameters in the
used diagram
in the from
diagram fromFigure
Figure16
16 according
according toto[39]
[39]are
are
asas follows:
follows:
√
66
IGI=  II DC (67) (67)
ππ DC
G

√
33 66
EDC=
EDC EEG ph (68) (68)
ππ G ph
66 X
RDC= 22 R
RDC Rss +
 2πX SS (69) (69)
2π
Commutation angle:
Commutation angle: r
2 XS ·IDC
τ = arccos( 1 − ) (70)
2 3XE  I DC
  arccos( 1  S G ph
) (70)
The shift angle between the generator EMF and 3 Ethe G ph generator current is defined by the

following The
relation:
shift angle between the generator EMFand the generator  current is defined by the following
relation:  τ − 12 sin(2τ) 
φ = arctan   (71)
sin 2
( τ )

   2 sin(2 ) 
1
  arctan   (71)
 sin ( ) 
2

The average value of the electromagnetic torque of a generator operating with a rectifier load
can be determined using the standard equation:

3 EG ph I G cos 
Tem(AV)   (72)
Ω
 and parameters of AFPM generators are summarized according to Table 1.

Table 1. Design AFPMG data.

Parameters and Dimensions of the Permanent Magnets of AFPM Generators

 5741
Energies 2020, 13, Magnets type: N40; Dimensions of a single magnet: 10 × 18 × 40 mm 18 of 28
 Br = 1.2 T; Hc = 899 kA/m; rm = 1.07;
   rs  = 0.0290 rad; a m = 18 mm; l m = 10 mm
The average value of the electromagnetic torque of a generator operating with a rectifier load can
 Number of magnets (on one rotor disc): 28; p = 14;
be determined using the standard equation:
 lc = 40 mm–for the construction with simple magnets (cases G1 and G3);
3E I cos φ
 l c = 70 mm–for skewedTmagnets=effective
Glength
ph G of the magnet (cases G2 and G4)
em(AV)− (72)
Construction of the stator in Ω
AFPM generator
 Tests
4. Laboratory R i =and mm; RVerification
270 Model o = 310 mm; rs = 290 mm;
'
 l c = 40 mm; a c = 50 mm; a sc = 30 mm; l  = 26 mm;
4.1. Characteristic
 l of=the Tested Generators
15 mm–for stator structure with cores
i


The verification of created
ws = 980; ; sc  rs  was
ps  7 models carried
= 0.1034   rfor
rad;out s
=generators,
0.1517 rad with the main elements in the
form of: two rotor
Rs =discs
2 Ω (each disc with a diameter of 650 mm and 28 permanent magnets placed
on the surface of one side), a stator with a diameter of 780 mm (with 21 coils, with non-overlapping
windings).TheThe laboratory
following namingequipment
conventionincluded a tested generator coupled by a torque measuring
was assumed:
shaft with a DC drive machine (Figures 17–19). Measurements were carried out for various resistive
 coreless generator with simple magnets-G1
loads andcoreless
a load generator
with a six-pulse diode
with skewed rectifier. The models’ verifications were extended by finite
magnets-G2
element
 analysis, performed in the ANSYS
generator with cores and simple magnets-G3 Maxwell environment (Figure 1). The main dimensions

and parameters of with
generator AFPM generators
cores and skewed aremagnets-G4
summarized according to Table 1.

Energies 2020, 13, x FOR PEER REVIEW 20 of 29

Figure 17. Laboratory
Figure17. set.
Laboratory set.

Figure 18. Stator disk.

Figure 18. Stator disk.

(a) (b)
Energies 2020, 13, 5741 19 of 28
Figure 18. Stator disk.

(a) (b)
Figure
Figure 19.19. Rotordiscs;
Rotor discs;(a)
(a)simple
simple magnets;
magnets;(b)
(b)skewed magnets.
skewed magnets.

4.2. Models Verification Table 1. Design AFPMG data.

Parameters
4.2.1. Verification of theand Dimensions
Induced EMFs of the Permanent Magnets of AFPM Generators

• TheMagnetsmaintype:goal N40;
of performed
Dimensions analysis of selected
of a single magnet: AFPM
10 × 18generator
× 40 mm models was to find the
construction
• Br = 1.2variants for which
T; Hc = 899 kA/m; itµrm
is possible
= 1.07; to increase the obtained generator power while keeping
the
• proper
β ( rs ) =machine work.
0.0290 rad; am = 18 mm; lm = 10 mm
• TheNumber first of
step of created
magnets (on onemodel 28; p = was
verification
rotor disc): 14; the comparison of induced generators’
• lc = 40 mm–for
electromotive force EMF. Figures 20–23
the construction show
with the EMF
simple waveforms
magnets forand
(cases G1 the first
G3); phase of the generator
for lc = 70 mm–for
• rotational speed skewed
of 206 rpm (48 Hz),
magnets obtained
effective lengthfrom finite
of the element
magnet analyses-FEM
(cases G2 and G4) (performed in
ANSYS Maxwell) as well as from analytical calculations and laboratory tests of four generator
topologies (G1–G4). Construction of the stator in AFPM generator
Figure 24a–d show the induced EMF spectra presented in dB for the tested generator design
• Ri =(G1–G4)
variants 270 mm;for = 310 mm;
Rorotational rs = 290
speed of 206
mm; rpm (48 Hz). The reference value was assumed to be 1
0
mV.
• lc = 40 mm; ac = 50 mm; asc = 30 mm; lδ = 26 mm;
li = 15 mm–for
• Figures 20–24 show
statorthe satisfactory
structure compatibility of the obtained results. For all cases, the
with cores
• ws = 980; ps = 7; αsc (rs ) = 0.1034 rad; ε(10%.
difference in voltage RMS values is less than rs ) =In case of
0.1517 radFFT spectrum of the electromotive force,
the
• Rs = 2 Ω
fundamental harmonic is consistent. The THD factors being an indicator of the content of higher
harmonics for EMF (THDE) are presented in Table 2.

The following naming convention was assumed:

• coreless generator with simple magnets-G1

• coreless generator with skewed magnets-G2
• generator with cores and simple magnets-G3
• generator with cores and skewed magnets-G4

4.2. Models Verification

4.2.1. Verification of the Induced EMFs

The main goal of performed analysis of selected AFPM generator models was to find the
construction variants for which it is possible to increase the obtained generator power while keeping
the proper machine work.
The first step of created model verification was the comparison of induced generators’ electromotive
force EMF. Figures 20–23 show the EMF waveforms for the first phase of the generator for rotational
speed of 206 rpm (48 Hz), obtained from finite element analyses-FEM (performed in ANSYS Maxwell)
as well as from analytical calculations and laboratory tests of four generator topologies (G1–G4).
Energies
Energies
Energies 13,
2020, 2020,
Energies 5741
2020,
2020, 13,xxxFOR
13,
13, FORPEER
FOR PEERREVIEW
PEER REVIEW
REVIEW 21of
21
21 of 20 of 28
of29
29
29
Energies 2020, 13, x FOR PEER REVIEW 21 of 29

(a)
(a)
(a) (b)
(b)
(b) (c)
(c)
(c)
(a) (b) (c)
Figure20.
Figure
Figure 20. Waveform
20. Waveform of
Waveform of induced
of induced electromotive
induced electromotive force
electromotive force (EMF)
force (EMF) (phase
(EMF) (phase1)
(phase 1) for
1) forthe
for the basic
the basicmodel
basic model G1:
model G1:(a)
G1: (a)
(a)
numerical
numerical
Figure 20.
numerical calculations
calculations
Waveform of
calculations finite-element
finite-element
induced method
method
electromotive
finite-element method (FEM);
(FEM);
force
(FEM);(EMF)(b)
(b)
(b) analytical
analytical
(phase 1)
analytical calculations;
calculations;
for the basic
calculations;
Figure 20. Waveform of induced electromotive force (EMF) (phase 1) for the basic model G1: (a) numerical (c)
(c)
model
(c) laboratory
laboratory
G1: (a)
laboratory
measurements.
measurements.
numerical calculations method
finite-element method (FEM); calculations;
(b) analytical (c)calculations;
measurements.
calculations finite-element (FEM); (b) analytical laboratory(c)measurements.
laboratory
measurements.

(a)
(a)
(a) (b)
(b)
(b) (c)
(c)
(c)
(a) (b) (c)
Figure
Figure
Figure
Figure 21.Waveform
21.
21. Waveform
Waveform
21. Waveform ofinduced
of
of induced
induced
of induced EMF EMF
EMF
EMF (phase1)
(phase
(phase
(phase 1)for
1)
1) forthe
for
for theoblique
the
the obliquemodel
oblique
oblique model
model
modelG2:
G2:
G2:
G2:(a)(a)
(a)
(a) numerical
numerical
numerical calculations
calculations
calculations
numerical calculations
FEM;
FEM;
Figure (b)
(b)
21. analytical
analytical
Waveform calculations;
calculations;
of induced (c)
(c)
EMF laboratory
laboratory
(phase 1) measurements.
measurements.
for the oblique model G2: (a) numerical calculations
FEM;FEM; (b) analytical
(b) analytical calculations;
calculations; (c)(c) laboratorymeasurements.
laboratory measurements.
FEM; (b) analytical calculations; (c) laboratory measurements.

(a)
(a)
(a) (b)
(b)
(b) (c)
(c)
(c)
(a) (b) (c)
FigureFigure
Figure
Figure 22.Waveform
22.
22. Waveform
Waveform
22. Waveform of ofofinduced
of induced
induced
induced EMF(phase
EMF
EMF
EMF (phase1)
(phase
(phase 1)for
1)
1) forthe
for
for thebasic
the
the basic
basic model
basicmodel
model
model with
with
with
withcores
cores
cores G3:G3:
G3:
G3:
cores (a)numerical
(a)
(a) numerical
numerical
(a) numerical
calculations
calculations
Figure 22.
calculations FEM;
FEM;
FEM; (b)
(b)
Waveform(b) analytical
analytical
of
analytical calculations;
calculations;
induced EMF (phase
calculations; (c)
(c)1)
(c) laboratory
laboratory
for the
laboratory
calculations FEM; (b) analytical calculations; (c) laboratory measurements. measurements.
measurements.
basic model with
measurements. cores G3: (a) numerical
calculations FEM; (b) analytical calculations; (c) laboratory measurements.

(a)
(a)
(a) (b)
(b)
(b) (c)
(c)
(c)
(a) (b) (c)
FigureFigure
Figure 23.Waveform
23. Waveform
23. Waveform
Figure 23. Waveform ofinduced
of induced
of induced
of inducedEMFEMF
EMF
EMF (phase1)
(phase
(phase
(phase 1)for
1)
1) forthe
for
for theoblique
the
the obliquemodel
oblique
oblique model
model
model with
with
with
with cores
cores G4:G4:
G4:
cores
cores G4: (a)numerical
(a)
(a) numerical
(a) numerical
numerical
calculations
calculations
Figure 23.
calculations
calculations FEM; FEM;
FEM;
FEM; (b)analytical
(b) (b)
Waveform (b) analytical
of induced
analytical
analytical calculations;
calculations;
EMF (phase
calculations;
calculations; (c)
(c) laboratory
laboratory
1)laboratory
(c)
(c) for measurements.
measurements.
the oblique
laboratory model with cores G4: (a) numerical
measurements.
measurements.
calculations FEM; (b) analytical calculations; (c) laboratory measurements.
Figure 24a–d show the induced EMF spectra presented in dB for the tested generator design
variants (G1–G4) for rotational speed of 206 rpm (48 Hz). The reference value was assumed to be 1 mV.
Energies 2020, 13, 5741 21 of 28
Energies 2020, 13, x FOR PEER REVIEW 22 of 29

(a) (b)

(c) (d)
Figure
Figure 24. FFT
24. FFT spectrum
spectrum of EMF
of EMF forforlaboratory
laboratory measurements,
measurements, finite
finiteelement analyses
element andand
analyses analytical
analytical
calculations: (a) generator G1; (b) generator G2; (c) generator G3; (d) generator
calculations: (a) generator G1; (b) generator G2; (c) generator G3; (d) generator G4. G4.

Table
Figures 2. Comparison
20–24 of the resultscompatibility
show the satisfactory obtained from of
analytical modelsresults.
the obtained and laboratory tests for
For all cases, THD
the difference in
factors of EMF.
voltage RMS values is less than 10%. In case of FFT spectrum of the electromotive force, the fundamental
harmonic is consistent. The THD factors being an indicator of the content THDof Ehigher harmonics for EMF
AFPMG
(THDE ) are presented in Table 2. Analytical Calculations Measure
G1 Single magnets; coreless stator 6.1% 6.5%
G2 Oblique magnets; coreless stator 2.0%
Table 2. Comparison of the results obtained from analytical models and laboratory tests 2.2%for THD
factorsG3of EMF. Single magnets; stator with cores 6.0% 7.3%
G4 Oblique magnets; stator with cores 1.3% 1.9%
THDE
AFPMG
The differences in higher harmonics can be considered acceptable,
Analytical due to the technological
Calculations Measure
aspects of creating a laboratory model (e.g., the difficulty of setting the constant air gap) and the
G1 Single magnets; coreless stator 6.1% 6.5%
accuracy of numerical calculations (FEM), depending on the model projection and mesh settings.
G2 Oblique magnets; coreless stator 2.0% 2.2%
G3 Single magnets; stator with cores 6.0% 7.3%
4.2.2. Verification of Operating Simplified Models
G4 Oblique magnets; stator with cores 1.3% 1.9%
The verification of simplified analytical models was based on the performed laboratory tests.
The obtained results are presented in Figures 25–32. Selected generator characteristics are presented
The differences in higher harmonics can be considered acceptable, due to the technological aspects
for different rotational speed values for four generator topologies (G1–G4). The figures present the
of creating a laboratory
following waveformsmodel (e.g.,resistive
at selected the difficulty
loads: of setting power
generator the constant air gap)
as a function and the and
of rotation accuracy
the of
numerical calculations (FEM), depending on the model projection and mesh settings.
AFPM generator external characteristics (UG = f(IG)) as well as the characteristics presenting the
waveforms for the case when generator operates with a diode rectifier with a resistive load (for G1
4.2.2. RVerification of Operating Simplified Models
L = 20 Ω, for G2 RL = 40 Ω, for G3 and for G4 RL = 80 Ω).

The verification of simplified analytical models was based on the performed laboratory tests.
The obtained results are presented in Figures 25–32. Selected generator characteristics are presented
for different rotational speed values for four generator topologies (G1–G4). The figures present the
following waveforms at selected resistive loads: generator power as a function of rotation and the AFPM
Energies 2020, 13, 5741 22 of 28

generator external characteristics (UG = f(IG )) as well as the characteristics presenting the waveforms
for the case when generator operates with a diode rectifier with a resistive load (for G1 RL = 20 Ω,
RL =2020,
for G2Energies 40 Ω, for
13, x G3
FOR and
PEER for G4 RL = 80 Ω).
REVIEW 23 of 29
Energies
Energies 2020,
2020, 13,
13, xx FOR
FOR PEER
PEER REVIEW
REVIEW 23
23 of
of 29
29

• 
Coreless

generator
Coreless
Coreless with
generator
generator simple
with
with magnets–G1
simple
simple magnets–G1
magnets–G1
Coreless generator with simple magnets–G1

(a)
(a) (b)
(b)
(a) (b)
Figure 25.
Figure 25. (a)
(a) Generator G1-power
G1-power as as a function ofof rotational speed
speed for
for different resistive
resistive loads;
loads; (b)
(b)
Figure 25. (a)
Figure 25. (a) Generator
Generator G1-power
Generator G1-power as aa function
as a function
function of rotational
of rotational speedspeed
rotational for different
for different resistive
different loads;loads;
resistive (b) generator
(b)
generator G1-voltage
generator G1-voltage as as aa function
function of of current
current forfor selected
selected rotational
rotational speed
speed values;
values; points-
points-
generator
G1-voltage as aG1-voltage
function ofascurrent
a function of current
for selected for selected
rotational rotational
speed values; speed values; points-
points-measurement results,
measurement results,
results, solid
solid line-analytical calculations.
calculations.
solid measurement
measurement
line-analytical results, solid line-analytical
calculations.line-analytical calculations.

(a)
(a) (b)
(b)
(a) (b)
Figure 26. Generator G1: (a) voltage changes with the rectifier load; (b) electric power changes with
Figure 26.
FigureFigure 26. Generator
Generator
26. Generator G1:
G1: (a) (a)
(a) voltage
G1:voltage
voltage changes
withwith
changes
changes with the
the rectifier
rectifier
the rectifier load;load;
load; (b)
(b) electric
electric
(b) electric power
powerpower changes
changes
changes with
with
with rectifier
rectifier load;
rectifier load; for
for selected
selected rotational
rotational speed
speed values,
values, points-measurement
points-measurement results,
results, solid
solid line-analytical
line-analytical
load; rectifier
for load;rotational
selected
calculations.
for selected rotational
speed speed
values, values, points-measurement
points-measurement results, results,
solid solid line-analytical
line-analytical calculations.
calculations.
calculations.
• Coreless
 generator
Coreless with
generator skewed
with magnets-G2
skewed magnets-G2
 Coreless
Coreless generator
generator with
with skewed
skewed magnets-G2
magnets-G2

(a)
(a) (b)
(b)
(a) (b)
FigureFigure
Figure 27.
27. (a)
27. (a) Generator
Generator
(a) G2-power
G2-power as a function
as a function of rotational
of rotational speed
speedspeed for different
for different resistive
resistive loads;
loads;loads; (b)
(b) generator
Figure 27. (a) Generator
Generator G2-power
G2-power as
as aa function
function of
of rotational
rotational speed for
for different
different resistive
resistive loads; (b)
(b)
generator
generator
G2-voltage as a G2-voltage
G2-voltage
function ofas
as a function
a
currentfunction
for of current
of current
selected for selected
for
rotational selected
speed rotational
rotational
values; speed values;
speed values; points-
points-measurement points-
results,
generator G2-voltage as a function of current for selected rotational speed values; points-
measurement results,
results, solid
solid line-analytical
line-analytical calculations.
calculations.
solid measurement
line-analytical calculations.
measurement results, solid line-analytical calculations.
Energies 13, 5741
2020, 2020,
Energies 13, x FOR PEER REVIEW 23 of 28
24 of 29
Energies 2020, 13, x FOR PEER REVIEW 24 of 29
Energies 2020, 13, x FOR PEER REVIEW 24 of 29

(a) (b)
(a) (b)
(a) (b)
Figure 28. Generator G2: (a) voltage changes with the rectifier load; (b) electric power changes with
Figure 28. Generator G2: (a) voltage changes with the rectifier load; (b) electric power changes with
FigureFigure
rectifier 28. Generator
load;
28. Generator for G2:voltage
(a) voltage
G2:selected
(a) changes
rotational
changes withwith
values, the rectifier
load;load;
thepoints-measurement
rectifier (b)results,
electric
(b) electric power power
solid changes
withwith
line-analytical
changes rectifier
rectifier load; for selected rotational values, points-measurement results, solid line-analytical
rectifier load;
calculations.
load; calculations. for selected rotational values, points-measurement results, solid
for selected rotational values, points-measurement results, solid line-analytical calculations. line-analytical
calculations.
 Core generator with simple magnets-G3
• 
Core generator
Core with
generator simple
with magnets-G3
simple magnets-G3
 Core generator with simple magnets-G3

(a) (b)
(a) (b)
(a) (b)
Figure 29. (a) Generator G3-power as a function of rotational speed for different resistive loads; (b)
FigureFigure
29. (a)29. (a) Generator
Generator G3-power
G3-power as a function
as a function of rotational
of rotational speedspeed for different
for different resistive
resistive loads;loads; (b)
(b) generator
Figure 29. G3-voltage
generator (a) GeneratorasG3-power
a functionas aoffunction
currentof for
rotational
selectedspeed for different
rotational speedresistive
values;loads; (b)
points-
generator
G3-voltage G3-voltage
as a function as
ofas a function
current of current
for selected for selected rotational speed values;
rotational speed values; points-measurement results, points-
generator G3-voltage
measurement results, solid a function ofcalculations.
line-analytical current for selected rotational speed values; points-
solid measurement
line-analytical results, solid line-analytical calculations.
calculations.
measurement results, solid line-analytical calculations.

(a) (b)
(a) (b)
(a) (b)
FigureFigure
30. 30. Generator
Generator G3: G3:voltage
(a) (a) voltage changes
changes withwith
the the rectifier
rectifier load;load;
(b) (b) electric
electric power
power changes
changes withwith
rectifier
Figure 30. Generator G3: (a) voltage changes with the rectifier load; (b) electric power changes with
Figure
load; rectifier
for 30. Generator
load;
selected G3:values,
for selected
rotational (a) voltage changes
rotational with
values, the rectifier
points-measurement
points-measurement results, load; (b)results,
solid electricsolid
powerline-analytical
line-analytical changes with
calculations.
rectifier load; for selected rotational values, points-measurement results, solid line-analytical
rectifier load; for selected rotational values, points-measurement results, solid line-analytical
calculations.
calculations.
calculations.
Energies 2020, 13, 5741 24 of 28

Energies
Energies 2020,
2020, 13,
13, xx FOR
FOR PEER
PEER REVIEW
REVIEW 25
25 of
of 29
29
• Core generator with skewed magnets-G4
 Core
Core generator
generator with
with skewed
skewed magnets-G4
magnets-G4

(a)
(a) (b)
(b)
FigureFigure
31. (a)31.
Figure 31. (a)
(a) Generator
Generator
Generator G4-power
G4-power
G4-power as
as aa function
as a function of
of rotational
of rotational
function speedspeed
rotational for
for different
for different
speed resistive
resistive
different loads;loads;
resistive (b)
(b) generator
loads; (b)
generator
G4-voltage as aG4-voltage
generator function ofas
G4-voltage as aa function
currentfunction of
of current
for selectedcurrent for
for selected
rotational selected rotational
rotational
speed values; speed
speed values;
values; points-
points-measurement points-
results,
measurement
measurement results,
results, solid
solid
solid line-analytical calculations. line-analytical
line-analytical calculations.
calculations.

(a)
(a) (b)
(b)
Figure 32.
32. Generator G4: (a)
(a) voltage changes
FigureFigure Generator
32. Generator G4:voltage
G4: (a) voltage
changes withwith
changes with the
the rectifier
load;load;
rectifier
the rectifier load; (b)
(b) electric
electric
(b) electric powerpower
power changes
withwith
changes
changes with
rectifier
rectifier
rectifier load;
load; for
for selected
selected rotational
rotational values,
values, points-measurement
points-measurement results,
results, solid
solid line-analytical
line-analytical
load; for selected rotational values, points-measurement results, solid line-analytical calculations.
calculations.
calculations.
The characteristics presented in Figures 25 and 26 confirm the satisfactory compliance of the
The
The characteristics
characteristics presented
presented in in Figures
Figures 25 25 and
and 26 26 confirm
confirm the the satisfactory
satisfactory compliance
compliance of of the
the
results for G1 obtained
results from the analytical models and from laboratory measurements, which confirms
results forfor G1 G1 obtained
obtained from from thethe analytical
analytical models
models and and fromfrom laboratory
laboratory measurements,
measurements, which which
the correctness
confirms
confirms the
ofcorrectness
the created
the correctness of models
of the
the created
formodels
the AFPMG
created models for
for the
coreless
the AFPMG
AFPMG coreless
generator
coreless generator
with simple
generator with
magnets.
with simple
simple
The differences
magnets.
magnets. The in the obtained
The differences
differences in results
in the
the obtained are less
obtained results than
results are 5%
are less for
less than the
than 5% resistive
5% for
for the load
the resistive of
resistive load the
load ofgenerator
of the and less
the generator
generator
than 10%
and for than
and less
less the
thanload 10% through
10% for
for the
the loadthethrough
load diode rectifier.
through the diodeBased
the diode rectifier.
rectifier. onBased
Figure
Based on27a
on it can
Figure
Figure 27abe
27a it seen
it can bethat
can be seen
seenfor a coreless
that
that for
for
aa coreless
structure with structure
coreless skewed with
structuredoublewith skewedmagnets
skewed double
double magnets
(G2), (G2),
(G2), it
it is possible
magnets it is
topossible
is obtain ato
possible obtain
topower
obtain of aa power
more of
power of more
than 4.5than
more kW, at a
than
4.5
4.5 kW,
generator kW, at
at aa generator
speed of 400 rpm
generator speed
(forof
speed G1,
of 400
400itrpm
was(for
rpm about
(for G1,
G1,2.1it
it was
kW).
was about
The2.1
about kW).
kW). The
generator
2.1 generator
generator characteristics
Thecharacteristics of Figures 27
characteristics
and 28 confirm the correctness of the analytical model for a coreless generator with skewed with
of
of Figures
Figures 27
27 and
and 28
28 confirm
confirm the
the correctness
correctness of
of the
the analytical
analytical model
model for
for a
a coreless
coreless generator
generator with
magnets.
skewed
skewed magnets.
magnets. The
The appearing
appearing differences
differences do
do not
not exceed
exceed 5%.
5%. For
For the
the G3
G3 generator,
generator, the
the convergence
convergence
The appearing differences do not exceed 5%. For the G3 generator, the convergence of the results with
of
of the
the results
results with
with resistive
resistive load
load while working
working with aa rectifier
rectifier cancan bebe considered
considered as correct
correct only for
resistive load while working with a while
rectifier can bewith considered as correct only foras100 rpm.onlyTheforresults
100
100 rpm.
rpm. TheThe results
results obtained
obtained from from analytical
analytical models
models and and laboratory
laboratory tests tests forfor rotational
rotational speed
speed higher
higher
obtained
than fromrpm analytical models and laboratory tests generator
for rotational speed higher than 150 rpm the slightly
than 150
150 rpm slightly
slightly differ
differ from
from each
each other.
other. For
For the
the generator with with skewed
skewed double double magnets
magnets (G4) (G4) the
differdifferences
from each
differences between
other.
between the
For the generator
the analytical
analytical model
with
model and
skewed
and laboratory
double
laboratory test
magnets
test were
were at at an
(G4)
an acceptable
the differences
acceptable level
level (less
between
(less than
than
the analytical
10%).
10%). ForFor model
the
the G3G3 andandG4
and G4laboratory
cases, test were
cases, relatively
relatively higher
higher atpowers
an acceptable
powers were
were obtainedlevel than
obtained (less
than for than
for G110%).
G1 and
and G2.
G2. For the G3 and
However,
However,
G4 cases,
for relatively
for both
both structures
structures higher(G3 powers
(G3 and G4),were
and G4), the obtainedcogging
the occurring
occurring than fortorque
cogging G1 and
torque was
was G2. at However,
at aa very highfor
very high both
level
level and
andstructures
the
the
vibrations
(G3 andvibrations of the whole structure were generated. The worst case in this regard was the case of the
G4), theof the whole
occurring structure
cogging were
torque generated.
was at aThe
very worsthigh case
levelin this
and regard
the was the
vibrations case
of of
the whole
the
G3.
structure This
G3. This fact
werefact caused
caused that
generated. that it was
it was
The worst practically
practically impossible
impossible
case in this regardfor for
was the
thetheG3 and
G3case
and of G4
G4theto operate
to operate
G3. This at a speed
at afact
speed of 200
of 200
caused that it
rpm
rpm andand higher.
was practically higher.
impossible for the G3 and G4 to operate at a speed of 200 rpm and higher.

4.3. Comparison of the Results Obtained from the Analytical Calculations and Laboratory Measurements
In order to prove and confirm correctness of developed models compare the analytical calculation
with real model measurements of PM disc generators, the results obtained for rotational speed of
206 rpm (48 Hz) for RMS phase values of EFM-EG (RMS) (zero current state) and current-IG (RMS)
Energies 2020, 13, 5741 25 of 28

(phase resistive load of 40 Ω) as well as the percentage differences between analytical calculations and
measurements related to measurements (∆EG (%) , ∆IG (%) ) are presented in Table 3.

Table 3. Comparison of the results obtained from analytical models and laboratory tests.

EG (RMS) IG (RMS)
AFPMG
Analytical Measure |∆EG (%) | Analytical Measure |∆IG (%) |
G1 Simple magnets; coreless stator 61.3 V 62.6 V 2.1% 1.65 A 1.69 A 2.4%
G2 Skewed magnets; coreless stator 101.5 V 105.1 V 3.4% 2.42 A 2.46 A 1.6%
G3 Simple magnets; stator with cores 101.3 V 95.8 V 5.7% 2.29 A 2.23 A 2.6%
G4 Skewed magnets; stator with cores 143.4 V 140.9 V 1.8% 3.52 A 3.43 A 2.6%

The results from Section 4.2 and Table 3 confirm that for the analytical and numerical calculations
and for the laboratory tests, the differences between the created models can be considered as acceptable.
According to Table 3 one can see a good results agreement in the form of RMS voltage and current,
obtained from the analytical calculations and the laboratory measurements for the four models (G1–G4).
It should also be mentioned that for the topology of the generator with cores considered in this
paper (where p = 14 and Zs = 3ps = 21) the cogging torque occurs, with the fundamental harmonic
equal to q = ±4Zs = ±12ps = ±6p = ±84. The next harmonics of the cogging torque will be multiples
of 84 (168, 252 . . . ). This information may be important for torque analysis and, above all, for further
search for optimal design solutions.

5. Conclusions
The paper presents the modelling methodology and test results for four design topologies of an
AFPM generator. The class of symmetrical three-phase AFPM disk generators with a symmetrical
structure was selected as representative, which, however, does not limit the possibility of developing
models using the presented methodology for machines with internal asymmetry of windings and
magnetic circuit.
In order to create the equations of mathematical models, the Lagrange formalism was used,
defining a modified co-energy function and the winding characteristics. The parameters of the created
models are integral values, so the results accuracy obtained based on the created circuit models of
generators excited by PM are limited. Some discrepancies occur because many phenomena occurring
in real laboratory models, are not and cannot be easily represented in discussed mathematical models
(for example: magnetic voltage drops, iron saturation, eddy currents).
In this paper, for most cases (for the following quantities: EMF, currents, voltages, powers), for the
performed calculation and measurement results the authors obtained the error level of less than 10%,
which can be considered a satisfactory result. This result confirms the assumption that the developed
circuit models of disc generators excited by PM are sufficiently accurate and can be useful for the
analysis of various operational issues and the study of their properties in the final stages of the design
process and also for diagnostic purposes.
The main reason for some discrepancies in the results is the inaccuracy and imprecision in
the assembling of the physical model, which was characterized by certain asymmetries due to
the unbalanced axial magnetic forces between the rotors. These forces twist the structure very
easily and make it difficult to set a constant regular length of air gap around the circumference.
Unfortunately, this is a disadvantageous feature of the presented AFPMG construction. In this case,
a more complex bearing arrangement and oversized, thicker rotor discs are required.
The performed calculations and measurement tests show that placing the iron cores inside the
stator windings resulted in an increase of the obtained power, with simultaneous formation of cogging
torques and torque pulsations. Compatibility of the created models has been confirmed for this solution.
On the other hand, for practical solutions (from a technological point of view), this variant requires
Energies 2020, 13, 5741 26 of 28

optimization, due to the difficulty of precisely setting the constant air gap and the occurrence of very
high vibration levels related to the cogging torques.
Summarizing, the main goal set by the authors was to present the AFPMG modelling methodology
using analytical methods, and the presented examples FEM calculations and laboratory tests confirmed
its correctness and usefulness.

Author Contributions: Conceptualization, N.R.-P. and T.W.; methodology N.R.-P. and T.W.; validation, N.R.-P.,
T.W. and D.B.; investigation, N.R.-P., T.W. and D.B.; data curation, N.R.-P. and T.W.; writing—original draft
preparation, N.R.-P. and T.W.; writing—review and editing, N.R.-P. and T.W.; supervision, N.R.-P. and T.W.;
All authors have read and agreed to the published version of the manuscript.
Funding: This work presented in this paper was funded by subsidies on science granted by Polish Ministry of
Science and Higher Education for Cracow University of Technology.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Chalmers, B.; Spooner, E. An axial-flux permanent-magnet generator for a gearless wind energy system.
IEEE Trans. Energy Convers. 1999, 14, 251–257. [CrossRef]
2. Caricchi, F.; Crescimbini, F.; Honrati, O. Modular axial-flux permanent-magnet motor for ship propulsion
drives. IEEE Trans. Energy Convers. 1999, 14, 673–679. [CrossRef]
3. Chen, Y.; Pillay, P.; Khan, A. PM Wind Generator Topologies. IEEE Trans. Ind. Appl. 2005, 41, 1619–1626.
[CrossRef]
4. Chan, T.F.; Lai, L.L. An Axial-Flux Permanent-Magnet Synchronous Generator for a Direct-Coupled
Wind-Turbine System. IEEE Trans. Energy Convers. 2007, 22, 86–94. [CrossRef]
5. Park, Y.-S.; Jang, S.-M.; Choi, J.-H.; Choi, J.-Y.; You, D.-J. Characteristic Analysis on Axial Flux Permanent
Magnet Synchronous Generator Considering Wind Turbine Characteristics According to Wind Speed for
Small-Scale Power Application. IEEE Trans. Magn. 2012, 48, 2937–2940. [CrossRef]
6. Borkowski, D.; Wegiel, T. Small Hydropower Plant with Integrated Turbine-Generators Working at Variable
Speed. IEEE Trans. Energy Convers. 2013, 28, 452–459. [CrossRef]
7. Kaňuch, J.; Ferkova, Z. Design and simulation of disk stepper motor with permanent magnets. Arch. Electr.
Eng. 2013, 62, 281–288. [CrossRef]
8. Gieras, J.F.; Wang, R.-J.; Kamper, M.J. Axial Flux Permanent Magnet Brushless Machines; Kluwer Academic
Publishers: Dordrecht, The Netherlands, 2004.
9. Tesla, N. Electro-Magnetic Motor. U.S. Patent 405,858, 3 December 1889.
10. Kirtley, J.L. Permanent Magnets in Electric Machines; John Wiley & Sons Ltd.: Chichester, UK, 2019; pp. 371–396.
11. Aydin, M.; Huang, S.; Lipo, T.A. Axial flux permanent magnet disc machines: A review, Wisconsin Electric
Machines & Power Electronics Consortium—Report. Conf. Rec. SPEEDAM 2004, 8, 61–71.
12. Kamper, M.J.; Wang, R.-J.; Rossouw, F. Analysis and Performance of Axial Flux Permanent-Magnet Machine
with Air-Cored Nonoverlapping Concentrated Stator Windings. IEEE Trans. Ind. Appl. 2008, 44, 1495–1504.
[CrossRef]
13. Zhilichev, Y.N. Three-dimensional analytic model of permanent magnet axial flux machine. IEEE Trans.
Magn. 1998, 34, 3897–3901. [CrossRef]
14. Parviainen, A.; Niemela, M.; Pyrhonen, J. Modeling of Axial Flux Permanent-Magnet Machines. IEEE Trans.
Ind. Appl. 2004, 40, 1333–1340. [CrossRef]
15. Azzouzi, J.; Barakat, G.; Dakyo, B. Quasi-3-D Analytical Modeling of the Magnetic Field of an Axial Flux
Permanent-Magnet Synchronous Machine. IEEE Trans. Energy Convers. 2005, 20, 746–752. [CrossRef]
16. Wang, R.-J.; Kamper, M.J.; Van Der Westhuizen, K.; Gieras, J. Optimal design of a coreless stator axial flux
permanent-magnet generator. IEEE Trans. Magn. 2005, 41, 55–64. [CrossRef]
17. Virtic, P.; Pisek, P.; Marcic, T.; Hadziselimovic, M.; Stumberger, B. Analytical Analysis of Magnetic Field and
Back Electromotive Force Calculation of an Axial-Flux Permanent Magnet Synchronous Generator with
Coreless Stator. IEEE Trans. Magn. 2008, 44, 4333–4336. [CrossRef]
18. Hosseini, S.; Agha-Mirsalim, M.; Mirzaei, M. Design, Prototyping, and Analysis of a Low Cost Axial-Flux
Coreless Permanent-Magnet Generator. IEEE Trans. Magn. 2007, 44, 75–80. [CrossRef]
Energies 2020, 13, 5741 27 of 28

19. Choi, J.-Y.; Lee, S.-H.; Ko, K.-J.; Jang, S.-M. Improved Analytical Model for Electromagnetic Analysis of Axial
Flux Machines with Double-Sided Permanent Magnet Rotor and Coreless Stator Windings. IEEE Trans.
Magn. 2011, 47, 2760–2763. [CrossRef]
20. Sung, S.-Y.; Jeong, J.-H.; Park, Y.-S.; Choi, J.-Y.; Jang, S.-M. Improved Analytical Modeling of Axial Flux
Machine with a Double-Sided Permanent Magnet Rotor and Slotless Stator Based on an Analytical Method.
IEEE Trans. Magn. 2012, 48, 2945–2948. [CrossRef]
21. Stamenkovic, I.; Milivojevic, N.; Schofield, N.; Krishnamurthy, M.; Emadi, A. Design, Analysis, and
Optimization of Ironless Stator Permanent Magnet Machines. IEEE Trans. Power Electron. 2012, 28, 2527–2538.
[CrossRef]
22. Jin, P.; Yuan, Y.; Minyi, J.; Shuhua, F.; Heyun, L.; Yang, H.; Ho, S.L. 3-D Analytical Magnetic Field Analysis of
Axial Flux Permanent-Magnet Machine. IEEE Trans. Magn. 2014, 50, 1–4. [CrossRef]
23. Huang, Y.; Zhou, T.; Dong, J.; Lin, H.; Yang, H.; Cheng, M. Magnetic Equivalent Circuit Modeling of
Yokeless Axial Flux Permanent Magnet Machine with Segmented Armature. IEEE Trans. Magn. 2014, 50, 1–4.
[CrossRef]
24. Maryam, S.; Naghi, R.; Vahid, B.; Juha, P.; Majid, R. Comparison of Performance Characteristics of Axial-Flux
Permanent-Magnet Synchronous Machine with Different Magnet Shapes. IEEE Trans. Magn. 2015, 51, 1–6.
25. Aydin, M.; Zhu, Z.Q.; Lipo, T.A.; Howe, D. Minimization of Cogging Torque in Axial-Flux Permanent-Magnet
Machines: Design Concepts. IEEE Trans. Magn. 2007, 43, 3614–3622. [CrossRef]
26. Tiegna, H.; Amara, Y.; Barakat, G. Study of Cogging Torque in Axial Flux Permanent Magnet Machines
Using an Analytical Model. IEEE Trans. Magn. 2014, 50, 845–848. [CrossRef]
27. Aydin, M.; Gulec, M. Reduction of Cogging Torque in Double-Rotor Axial-Flux Permanent-Magnet Disk
Motors: A Review of Cost-Effective Magnet-Skewing Techniques with Experimental Verification. IEEE Trans.
Ind. Electron. 2013, 61, 5025–5034. [CrossRef]
28. Wanjiku, J.; Khan, M.A.; Barendse, P.S.; Pillay, P. Influence of Slot Openings and Tooth Profile on Cogging
Torque in Axial-Flux PM Machines. IEEE Trans. Ind. Electron. 2015, 62, 7578–7589. [CrossRef]
29. Wegiel, T. Space harmonic interactions in axial flux permanent magnet generator. Tech. Trans. Electr. Eng.
2016, 65–79. [CrossRef]
30. Radwan-Pragłowska, N.; Borkowski, D.; Wegiel, T. Model of coreless axial flux permanent magnet generator.
In Proceedings of the 2017 International Symposium on Electrical Machines (SME), Naleczow, Poland,
18–21 June 2017; pp. 1–6. [CrossRef]
31. Radwan-Pragłowska, N.; W˛egiel, T.; Borkowski, D. Parameters identification of coreless axial flux permanent
magnet generator. Arch. Electr. Eng. 2018, 67, 391–402.
32. Radwan-Praglowska, N. Mathematical Model of PM Disc Generator with Iron Cores of Stator Coils.
In Proceedings of the 2018 14th Selected Issues of Electrical Engineering and Electronics (WZEE), Szczecin,
Poland, 19–21 November 2018; pp. 1–6. [CrossRef]
33. Heller, B.; Hamata, V. Harmonic Field Effects in Induction Machines; Elsevier: Amsterdam, The Netherlands,
1977.
34. Sobczyk, T.J.; Drozdowski, P. Inductances of electrical machine winding with a nonuniform air-gap. Electr. Eng.
1993, 76, 213–218. [CrossRef]
35. Wegiel, T. Space Harmonic Interactions in Permanent Magnet Generator; Monograph No 447; Cracow University
of Technology: Cracow, Poland, 2013.
36. Wegiel, T. Cogging torque analysis based on energy approach in surface-mounted PM machines.
In Proceedings of the 2017 International Symposium on Electrical Machines (SME), Naleczow, Poland,
18–21 June 2017; pp. 1–6. [CrossRef]
37. Di Gerlando, A.; Foglia, G.; Iacchetti, M.F.; Perini, R. Analysis and Test of Diode Rectifier Solutions in
Grid-Connected Wind Energy Conversion Systems Employing Modular Permanent-Magnet Synchronous
Generators. IEEE Trans. Ind. Electron. 2011, 59, 2135–2146. [CrossRef]
38. Sudhoff, S.; Corzine, K.; Hegner, H.; Delisle, D. Transient and dynamic average-value modeling of synchronous
machine fed load-commutated converters. IEEE Trans. Energy Convers. 1996, 11, 508–514. [CrossRef]
Energies 2020, 13, 5741 28 of 28

39. Borkowski, D. Average-value model of energy conversion system consisting of PMSG, diode bridge rectifier
and DPC-SVM controlled inverter. In Proceedings of the 2017 International Symposium on Electrical
Machines (SME), Naleczow, Poland, 18–21 June 2017; pp. 1–6. [CrossRef]

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional
affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).