machines
Article
Challenges in Winding Design and Thermal Calculations:
Physical Model of Permanent Magnet Synchronous Machine
Jan Laksar , Lukáš Veg *






Citation: Laksar, J.; Veg, L.;
Pechánek, R. Challenges in Winding
Design and Thermal Calculations:
Physical Model of Permanent Magnet
Synchronous Machine. Machines 2021,
9, 234. https://doi.org/10.3390/
machines9100234
Academic Editors: Antonio
J. Marques Cardoso, Payam
Shams Ghahfarokhi and
Amel Adouni
Received: 15 September 2021
Accepted: 10 October 2021
Published: 13 October 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 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 (https://
creativecommons.org/licenses/by/
4.0/).
Machines 2021, 9, 234. https://doi.org/10.3390/machines9100234
and Roman Pechánek
Research and Innovation Centre for Electrical Engineering, Faculty of Electrical Engineering, University of West
Bohemia, Univerzitní 26, 301 00 Pilsen, Czech Republic; laksar@fel.zcu.cz (J.L.); rpechane@fel.zcu.cz (R.P.)
* Correspondence: vegl@fel.zcu.cz
Abstract: Interest in multilayer windings is increasing with the application of the hairpin winding
technology to the manufacturing of electrical machines. Therefore, the four-layer fractional slot
concentrated winding is used for the initial design of the machine in this paper. The proposed
physical model of the machine uses winding with a relatively high number of turns which is
inappropriate to hairpin winding. Therefore the round-wire winding is created and the three-layer
winding is derived and analyzed including the effect on the slot leakage inductance. The thermal
analysis is then applied to the physical model of the machine to evaluate the slot-related thermal
properties of the slot and the whole machine. The measurement is compared with the finite element
analysis (FEA) and the equivalent slot thermal conductivity and heat transfer coefficients of the stator
and rotor are obtained.
Keywords: permanent magnet synchronous machine; physical model; three-layer winding; slot
leakage inductance; magnetomotive force; slot thermal conductivity; heat transfer coefficient
1. Introduction
Modern trends in the design of permanent magnet synchronous machines (PMSM)
are focusing on different parts of the machine with a similar goal; increase the efficiency or
the power density of the machine. Both quantities are crucial aspects of the machine design,
especially in the electric vehicle industry. An increase of the efficiency of the machine (or,
better, the reduction of the losses of the machine) will help to decrease the requirements
for the cooling system. The reduction of the losses is also an important assumption in
the increase of the power density; the density of the heat which has to be removed from
the machine area increases with the decreasing volume of the machine (or increasing the
power). One of the promising approaches consists of the usage of a double-stator Vernier
permanent magnet machine [1], but the mechanical construction of the machine can be a
limiting point. The overall influence of several input geometrical parameters on the final
design is discussed in [2] to minimize the total loss of high-speed motor and the cooling
surface optimization of the electric kart motor is analyzed in [3]. The modern trends of
increasing the power density of the whole system are moving towards the integration of
power electronics and motor into one compact device [4].
The increase of the power density can be generally achieved by three approaches. The
most significant results can be reached by the increase of the speed of the machine [5]. The
machine can be then directly connected to the transmission with a fixed or two-speed [6]
gear ratio to obtain the required output speed. This approach emphasizes the requirements
of the bearings, and often leads to the decrease of the saturation of the magnetic circuit to
reduce the magnetic circuit losses or to an improved cooling system to effectively dissipate
the produced heat [7].
The second category of increasing the power density is based on the usage of new
materials in the magnetic circuit, winding insulation, or permanent magnets, etc. Magnetic
materials with a high level of saturation and small loss density [8] enable the reduction
https://www.mdpi.com/journal/machines
Machines 2021, 9, 234 2 of 15

of the volume of the stator stack which can be supported by permanent magnets with a
high density of energy. Properties of insulating materials can be improved in different
ways; insulation systems with higher temperature class allow the increase of the maximal
temperature (up to extreme temperatures [9]). Insulation systems with high dielectric
strength enable the use of a thinner insulation layer for defined input voltage.
The third category consists of the application of the new technologies to the production
of the machine. It is directly linked with the previously mentioned approaches to the
power density increase, e.g., the composite insulation systems or the whole technology
of the insulation. A very interesting technology, especially in the automotive industry
applications is the hairpin winding technology [10,11]. Hairpin windings combine the
advantages of both rectangular-wire windings (high slot fill factor) often supported by
missing coil insulation, and round-wire coils (semi-closed slots reducing the negative slot
opening-related effects as the torque ripple and general decrease of the inductance caused
by the slot openings) [12]. The biggest technological and manufacturing advantage of the
hairpin windings lies in the automatization of the winding process, which can help reduce
the manufacturing costs. Every turn of the coil is formed by one hairpin; multiturn coils
are then assembled by the multilayer hairpin winding. It enables to create the multilayer
windings with different connections to reduce the higher harmonic components or the
differential leakage inductance. The number of layers (turns) is mostly limited to 6 or 8 to
not to increase the number of winding connections.
Every improvement of the power density has to be supported by a sufficient cooling
system. The thermal calculation is the most challenging part of the design of the electrical
machine, especially in the dynamic operation as discussed in [3,7]. The calculations are
often supported by mathematical models and finite element analysis (FEA). Definition of
the thermal model of the slot is the crucial part of the description of the heat dissipation. The
slot consists of different layers with different thermal properties [13]. The most complicated
situation appears in the round-wired coils with random placement of the conductors
in the slot area. The equivalent thermal parameters (thermal conductivity [14,15] heat
transfer coefficients [16,17]) are often historically defined as empirical values based on
the comparison of the calculation and measurement. The same methodology is also used
in [18,19].
The automatization of the winding process is not focused only on the hairpin windings
but can be used for every winding type. It opens up new possibilities in the winding design.
A physical model of the small permanent magnet synchronous machine was designed as
a multipurpose model referring to the part of the modern and challenging parts of the
complex design of the electrical machine. The design of the multilayer machine is inspired
by the three-layer fractional slot concentrated winding described in [20]. The precise
analytical calculation of the inductance helps to evaluate the operation of the machine
especially at high speed where the flux weakening operation is necessary [21]. The slot
leakage inductance of fractional slot concentrated machines is often the most significant
component of permanent magnet machines and, therefore, it is analyzed in this paper. A
different approach to the definition of the ideal turn ratio is applied and the mathematical
function describing the magnetomotive force waveform obtained is derived. The influence
of the winding design on the inductance of the machine is analyzed then. The second
part of the paper is focused on the thermal analysis of the machine and comparison of the
calculations and measurements of the physical model to get real information about the
heat transfer in the slot.
It turns out that the proper definition of the slot heat transfer coefficients is the crucial
part of the thermal design of modern electric machines that often use water-cooled housing
and the slot is the area of the dissipating of the heat generated in the winding. Although
algorithms to increase the slot fill factor have been investigated [22], the real machines
with rounded wires often have fill factor in ratio 0.35–0.45. Therefore, the experimental
validation of the slot heat transfer coefficient of this physical model can be an important
part of the complex thermal design of modern electric machinery.
 Machines 2021, 9, x FOR PEER REVIEW 3 of 15
nes 2021, 9, x FOR PEER REVIEW 3 of 15

machines with rounded wires often have fill factor in ratio 0.35–0.45. Therefore, the exper-
Machines 2021, 9, 234machines with rounded wires often have fill factor in ratio 0.35–0.45. Therefore, the exper- 3 of 15
imental
imental validation of validation
the slot heatof the slot heat
transfer transfer
coefficient of coefficient of this
this physical physical
model can bemodel
an can be an
important
important part part of thermal
of the complex the complex thermal
design design
of modern of modern
electric electric machinery.
machinery.
2.2.Design
Designof
2. Design of Three-Layer Three-Layer
Three-LayerWinding
ofWinding Winding
Initial
Initial
Initial universal universal double-layer
universal three-phase
double-layer three-phase
double-layerfractional
three-phase slotfractional
fractional
concentratedslot
slot concentrated
concentrated
winding withwinding
winding with
with
the
the number
number ofofslots
slotsper
perpole
poleand phase
and phase q = 1/2
q = is replaced
1/2 is by
replaced
the number of slots per pole and phase q = 1/2 is replaced by winding with q = 2/5 to reduce winding
by with
winding q = 2/5
with qto
= reduce
2/5 to
the parasitic
reduce the effects
parasitic such as
effects cogging
such as torque
cogging and
torque pulsation
and
the parasitic effects such as cogging torque and pulsation of induced voltage and output of induced
pulsation of voltage
induced and output
voltage and
current [23]. current
output [23].
current
Unfortunately, Unfortunately,
[23].winding
this this
Unfortunately, winding
generates generates
thisawinding
higher a higher
generates
amount amountamount
of aparasitic
higher of parasitic
space space
of parasitic
harmonic
space content
harmonic in comparison
content in with
comparison the
with initial
the winding.
initial
harmonic content in comparison with the initial winding. Therefore, the possibility of winding.Therefore,
Therefore,the possibility
the of
possibility
space
of harmonic
space harmonic content reduction
content
space harmonic content reduction is analyzed. reductionis analyzed.
is analyzed.
Four-layerwinding
Four-layer windingisisobtained
obtainedwhen when twoidenticalidenticalsetssetsofofdouble-layer
double-layer windings are
Four-layer winding is obtained when two identical two sets of double-layer windings are windings are
usedandand the
the sets
sets shift by yyss slots
shift by slots (mostly
(mostly by by one
one slot
slot toto avoid
avoid high
high fundamental space space
used and the sets shift by ys slots (mostly by one slot to avoid high fundamental fundamental
used space
harmonicsreduction).
harmonics reduction). Current
Current polarity
polarity in in the
the shifted
shifted winding
winding set set has
has toto be
be changed
changed to to not
not
harmonics reduction). Current polarity in the shifted winding set has to be changed to not
to get
to get the opposite
opposite direction
direction ofofcurrents
currents inside
inside oneoneslot if the
slot if shift
the ys isysan
shift is odd
an number.
odd number.The
to get the opposite direction of currents inside one slot if the shift ys is an odd number.
principle of theof four-layer winding definition is shown in Figure 1.
The principleThe principle
of the four-layer the four-layer
winding winding
definition definition
is shown is shown
in Figure 1. in Figure 1.

++
- ++

+
- -+ - -- ++ - - -+

+
++ -+ +
+
+-
--

+
-+ +
--

+
-
+-

+
--

++

+
- -

+
-
- - - - -
-

+ +
- - - -
- - -
+

+
--

+
+
- -
- - -

+ + + = =
-
-
+ +
-
+ + ++

++
-
+ +
-
- -
- -
++

+ - + -

++
- -
- -
- - -

+
-
+ +

+
- + - +

+
+
- -
+ + +
+
+ + + +
+
- +

+
-
+ +
+

+ + --

+
-+ --
+-

+
+
-- -+ +-

+
-

+
-+ --
+-

+
+

-+ +-
-
- - + -
- -
+

+
-
- -
+

- -

Figure 1. Principle of four-layer winding definition.

Figure 1. Principle of four-layer winding definition.
Figure 1. Principle of four-layer winding definition.

Every
Everyof
Every slot consists slot
slot consists
consists
three of
ofthree
three
coil sides coil
coilsides
belonging to belonging
sides phase to
belonging
one toone
and one phase
one coil and
phase andone
side onecoil
be- coilside
sidebe-
be-
longing
longing to
tothe
the other
otherwinding.
winding. This
Thisslot
slotarrangement
arrangement can
can be
bedivided
divided into
into
longing to the other winding. This slot arrangement can be divided into three single-layer three
threesingle-layer
single-layer
winding,
winding,
winding, as shown as
as shown2. in
shown
in Figure in Figure
Figure 2.
2.

++ +
+

++ - - - - + - -
+

+ + +
+

- - -- -
+

-
+

+
- - -
+

+ + +
+

--
+
+

-
- - -
+

- -
- - - -
+ +

+ +

- - - - -
+

+
+

- - - -

= = + + + +
-
+ + +
- -
+
++

+ + +
- -
- - -
- -
+
++

- - -
- -
+

- -
+

- + - + + - -
+

+
+

+ + +
- + + + -
+

+ + -- +
+

- -
+

-- -+ -
+- -
+

+
+
+

- -
+
+

-
-+ - - - - - -
+
+

- - + - + -

Figure 2. Decomposition of four-layer winding to three single-layer windings.

Figure 2. Decomposition of four-layer winding to three single-layer windings.
Figure 2. Decomposition of four-layer winding to three single-layer windings.
Identical configuration of every single-layer winding can be obtained by two different
Identical configuration
Identical windings:
configuration of everywinding
of every single-layer single-layer winding
can be canby
obtained betwo
obtained
differ-by two differ-
ent windings:
ent windings:(a) fractional slot winding q = 2/5 with 10 poles;
(a) fractional (a)slotfractional
(b) slot
integer slot
winding winding
q =winding
2/5 with q10q= =poles;
2 2/5 with
with 10 poles;
2 poles.
(b) integer
(b) integer slot winding slot winding
q = 2iswith q = 2 with 2 poles.
2 poles. by the produced step curve of magnetomotive force Fm and
This effect represented
This effect This effect
itsisspectral isby
analysis
represented represented
in Figure
the produced 3.byThe
the produced
curve of step
fundamental
step curve ofcomponent
harmonic
magnetomotive magnetomotive ν = 1force
force Fmisand Fm and
for integral
its spectral
slot winding
its spectral analysis analysis
in Figureand in Figure 3.
ν = 5fundamental
3. The The fundamental
for fractionalharmonic
slot winding. harmonic component is ν
component is ν = 1 for integral = 1 for integral
slot winding and ν = 5 for fractional
slot winding and ν = 5 for fractional slot winding. slot winding.
es 2021, 9, xMachines
FOR PEER2021, 9, 234
REVIEW 4 of 15 4 of 15

Figure 3.Figure 3. Magnetomotive

Magnetomotive force stepforce
curvestep
andcurve and its spectrum
its spectrum produced produced by single-layer
by single-layer winding q winding q =10 poles.
= 2/5 with
2/5 with 10 poles.
The magnetomotive force produced by integral slot windings can be mathematically
described by
The magnetomotive equations
force produced derived in [24].slot
by integral Thewindings
resultingcan
three-layer winding consists of three
be mathematically
single-layer
described by equations windings
derived shifted
in [24]. by one slot
The resulting pitch represented
three-layer by angle
winding consists s. Windings can be
of ξthree
divided shifted
single-layer windings into two bytypes; onepitch
one slot central winding with
represented Ncc turns
by angle per one coil
ξs. Windings canand be two marginal
windings each with N
divided into two types; one central winding
cm turns with Ncc turns per one coil and two marginal of these three
per coil. The produced magnetomotive force
single-layer windings is described
windings each with Ncm turns per coil. The produced as a sum
magnetomotive force of these three
single-layer windings is described as a sum
∞
 
2 
( m −1) π
∞ Fm (ξ )2 = ∑ Fm1ν ∑ Nci sin (ωt ± ν(ξ + (i − 1)ξ s )) ∓ (ν ± 1) m ,
ν =1 i =0 (𝑚 − 1)𝜋 (1)
𝐹𝑚 (𝜉) = ∑ 𝐹𝑚1𝜈 [∑ 𝑁𝑐𝑖 sin ((𝜔𝑡 ± N 𝜈(𝜉 =+ (𝑖
N − 1)𝜉
= N )) ∓ (𝜈
𝑠 , N = N± 1) )] ,
c0 c2 cm c1 cc 𝑚 (1)
𝜈=1 𝑖=0
where ξ is𝑁the
𝑐0 = 𝑁𝑐2 =angular
space 𝑁𝑐𝑚 , 𝑁position,
𝑐1 = 𝑁𝑐𝑐 ωt the time angular position, m is the number of

where ξ is thephases and Fm1νposition,

space angular is the magnitude
ωt the timeof produced magnetomotive
angular position, m is the force
numberdefined
of in [24]. The
phases and Fm1ν is the magnitude of produced magnetomotive force defined in [24]. The description of
expression in brackets in (1) can be modified and the final mathematical
magnetomotive
expression in brackets force
in (1) can be produced
modified byandthree-layer winding is description of
the final mathematical
magnetomotive force produced by three-layer∞ winding is
( m − 1) π
 
∞ Fm ( ξ ) = ∑ 3Nc k csν Fm1ν · sin ( ωt ± νξ ) ∓ ( ν ± 1) , (2)
ν =1
(𝑚 − 1)𝜋 m
𝐹𝑚 (𝜉) = ∑ 3𝑁𝑐 𝑘𝑐𝑠𝜈 𝐹𝑚1𝜈 ∙ sin ((𝜔𝑡 ± 𝜈𝜉) ∓ ± 1) (𝜈 ) , (2)
𝑚
where Nc 𝜈=1
is the mean number of winding turns defined as Nc = (Ncc +2Ncm )/3 and kcsν is a
where Nc is thefactor
meanrespecting
number of thewinding
numberturnsof turns of coils
defined as and
Nc =shift of winding
(Ncc+2N cm)/3 andsets
kcsν equal
is a to
factor respecting the number of turns of coils and shift of winding sets equal to
2Ncm
csν = 1 −
2𝑁k𝑐𝑚 (1 + cos(νξ s )) (3)
𝑘𝑐𝑠𝜈 = 1 − 3Nc ))
(1 + cos(𝜈𝜉 (3)
𝑠
3𝑁𝑐
Preserving the average number of coil turns Nc constant, distribution of turns into
Preservingcentral
the average number coils
and marginal of coil
is turns Nc by
limited constant, distribution
two extreme of turns
situations; when intomarginal coils
central and marginal
windingscoils
areis not
limited
used,byratio
two2Nextreme situations; when marginal coils wind-
cm /3Nc equals zero. When winding with central coils is not
ings are not used, ratio 2N cm/3Nc equals zero. When winding with central coils is not used,
used, the ratio 2Ncm /3Nc equals one.
the ratio 2Ncm/3Nc equals one.
Integral slot winding produces only spatial harmonic component orders
Integral slot winding produces only spatial harmonic component orders
ν = 2km ∓ 1, k = 0, 1, 2, . . . (4)
𝜈 = 2𝑘𝑚 ∓ 1, 𝑘 = 0, 1, 2, … (4)
which can
which can be divided betwo
into divided intobased
groups two groups
on the based
relationonbetween
the relation between
factor distri-kcs and distribu-
kcs andfactor
bution of coils tion of The
turns. coilsinfluence
turns. The
of influence of turns
the ratio of the ratio of turns
to factor factor kin
toshown
kcs is is shown
cs Figure 4. in Figure 4.
 Machines2021,
Machines 2021,9,9,x 234
FOR PEER REVIEW 5 of1515
5 of

Figure
Figure4.4.Influence
Influenceofofturns
turnsratio
ratioon
onfactor
factorkcs.
kcs .

ItItisisobvious,
obvious,that
thatharmonic
harmoniccomponent
componentorders km±±
ordersνν==44km 1 can
1 can bebesuppressed
suppressedwhen
when
2𝑁cm
2N 𝑐𝑚
11−− 3𝑁 ((1
1+ cosξ𝜉s𝑠)) =
+cos = 00 (5)
(5)
3Nc𝑐
that
thatis,is,for,
for,
2N √ .
2𝑁cm
𝑐𝑚 = 4 − 2 3 = 0.54 (6)
3N = 4 − 2√3 =̇ 0.54 (6)
3𝑁c 𝑐
The ratio between the number of turns of the central and marginal coil is then
The ratio between the number of turns of the central and marginal coil is then
   √ 
2𝑁cc𝑐𝑐 2 1 − 2N
2N 2𝑁
cm𝑐𝑚
2 1 − 4 − 2 3 √
N𝑁cc 2 (1 − 3N )
𝑐 = 2(1 − 4 √ − 2√3) = 3
3Nc c
𝑐𝑐 = 3𝑁 𝑐= 3𝑁 (7)
N𝑁cm = 2N cm = 2N cm = 44−−22√3 3 = √3 (7)
𝑐𝑚 2𝑁
3Nc𝑐𝑚 2𝑁c𝑐𝑚
3N
3𝑁𝑐 3𝑁𝑐
which corresponds to the results in [20]. The presented method allows determining of
which corresponds to the results in [20]. The presented method allows determining of
winding factor for any ratio of central and marginal coil turns. The winding factor of the
winding factorharmonic
fundamental for any ratio of central(νand
component = 5)marginal coil turns.
derived from The winding
the initial factor
integral slot of the
two-pole
fundamental harmonic component
single layer winding is k’w5 = 0.966. (ν = 5) derived from the initial integral slot two-pole
single The
layerfactor
winding is k’w5 =the
respecting 0.966.
number of turns of coils and shift of winding sets is for the
The
ideal factor
ratio respecting
derived the number
in (6) and (7): of turns of coils and shift of winding sets is for the
ideal ratio derived in (6) and (7):
 √  
2π

k cs5 = 1 − 4 − 2 3 1 + cos 5· 2𝜋 = 0.928 (8)
𝑘𝑐𝑠5 = 1 − (4 − 2√3) (1 + 𝑐𝑜𝑠 (5 ∙12 )) = 0.928 (8)
12
The
Thewinding
windingfactor
factorofofoptimized
optimizedthree-layer
three-layerwinding
windingisisthen
then

𝑘k𝑤5 = 𝑘 ′0 𝑘k𝑐𝑠5 = 0.966 ∙ 0.928 = 0.897 (9)

w5 = k𝑤5w5 cs5 = 0.966·0.928 = 0.897 (9)
The coefficient of differential leakage inductance τdiff is then reduced from the value
The coefficient of differential leakage inductance τdiff is then reduced from the value
0.97 (two-layer winding) to 0.84.
0.97 (two-layer winding) to 0.84.
2.1.
2.1.Slot
SlotLeakage
LeakageInductance
Inductance
Coil
Coil sides insome
sides in someslots
slotsbelong
belongtotodifferent
differentphases
phasesififcoils
coilswith
withshort
shortpitch
pitchare
areused
used
for
for the multilayer windings. The total energy of the magnetic field stored in theseslots
the multilayer windings. The total energy of the magnetic field stored in these slots
decreases
decreasesand andalso
alsothetheslot
slotleakage
leakageinductance
inductanceLLσsσs . This
. Thisinductance
inductancereduction
reductionisisrespected
respected
bybycoefficients
coefficientskkCuCufor
forthe
theactive
activeslot
slotarea
areaand
andkkekkefor
forthe
theslot
slotopening
openingarea.
area.These
Thesecoefficients
coefficients
are
arederived
derivedinin[25]
[25]forfordouble-layer
double-layerwinding
windingand andare arewidely
widelyusedusedininthe
theliterature.
literature.
The
Theleakage
leakageinductance
inductanceininthe theslot
slotdepends
dependson onthethenumber
numberofofcoilcoilsides
sidesbelonging
belongingtoto
one phase and the placement of these coil sides in the slot. Possible
one phase and the placement of these coil sides in the slot. Possible coil side placementscoil side placementsin
inthe
theslot
slotare
aresummarized
summarizedininFigure Figure5.5.
 hs 2hc

Machines
h0
Machines2021,
2021,9,9,x234
FOR PEER REVIEW 6 6ofof15
15
bs (a) (b) (c) (d)

Figure 5. Overview of possible coil sides belonging to one phase placements in one slot: central
and one bm
bc marginal coil (a,b), and only marginal coil (c,d).

Coil widths bc and bm are defined by total slot width bs and number of turns Nc and
Nm as 2hc
hs 𝑁𝑐𝑐
𝑁𝑐𝑚
𝑏𝑐 = 𝑏𝑠
𝑁
h0 2 + 𝑐𝑐
𝑁𝑐𝑚
bs (a) (b) (c) (d) (10)
2
𝑏𝑚 = 𝑏𝑠
Overviewofofpossible
Figure5.5.Overview
Figure possiblecoil
coilsides
sidesbelonging
belonging 𝑁 phase
to one placements in one slot: central and
2to+one𝑐𝑐phase placements in one slot: central
oneone
and marginal coilcoil
marginal (a,b), andand
(a,b), onlyonly
marginal coil coil
marginal (c,d).𝑁𝑐𝑚
(c,d).
The analytical integral solution of magnetic energy stored in the slot is evaluated for
widthsbbc cand
Coilwidths andbbmmare
aredefined
defined bytotaltotalslot widthbbs sand
slotwidth andnumber
numberof ofturns
turns N Nccand
and
all Coil
combinations in Figure 5. The totalby magnetic energy produced by analyzed winding is
NNmcompared
masas
with theoretical full-slot winding
Ncc and the dependency of coefficients k Cu and k ke
Ncm 𝑁 𝑐𝑐 2
for different Ncc/Ncm ratios isbshown
c = in Figure
bs bm 6.= b (10)
2 +𝑏 NNcm
cc 𝑁𝑐𝑚 2 + Ncc s
Similar approach can be generalized 𝑐 = to different
𝑁
𝑏𝑠 Nnumber
cm of coil sides in one slot; the
arrangement of the coil sides belonging 2 +one𝑐𝑐
to phase stored
in the inslot has toisbe summarized
The analytical integral solution of magnetic 𝑁𝑐𝑚 energy the slot evaluated for
(according to Figure 5) and ratio of the
all combinations in Figure 5. The total magnetic energy of magnetic
energy field
produced respecting
by analyzedthe coil (10)is
position
winding
2
to theoretical
compared with full-slot winding
theoretical full-slotdefined
𝑏𝑚 =theand
winding coefficients kCu andofkkecoefficients
the 𝑏dependency . A typicalkexample
Cu and kke
of
𝑁𝑐𝑐 𝑠
multilayer
for different windings are the
Ncc /Ncm ratios is hairpin
shown in winding2 + analyzed
Figure 6. e.g., in [10].
𝑁𝑐𝑚
The analytical integral solution of magnetic energy stored in the slot is evaluated for
all combinations in Figure 5. The total magnetic energy produced by analyzed winding is
compared with theoretical full-slot winding and the dependency of coefficients kCu and kke
for different Ncc/Ncm ratios is shown in Figure 6.
Similar approach can be generalized to different number of coil sides in one slot; the
arrangement of the coil sides belonging to one phase in the slot has to be summarized
(according to Figure 5) and ratio of the energy of magnetic field respecting the coil position
to theoretical full-slot winding defined the coefficients kCu and kke. A typical example of
multilayer windings are the hairpin winding analyzed e.g., in [10].

Figure6.6.Dependency
Figure Dependencyofofcoefficients
coefficientskCu
kCuand
andkkkeke on
on the
the turns
turns ratio.
ratio.

2.2. Similar
Physicalapproach
Model of the
canPermanent Magnet
be generalized toSynchronous Machines
different number (PMSM)
of coil with
sides in oneThree-
slot; the
Layer Winding
arrangement of the coil sides belonging to one phase in the slot has to be summarized
(according
A small stator 5)
to Figure and with
stack ratio the
of the energy of shown
dimensions magneticin field
Figurerespecting
7 is usedthe
forcoil
theposition
physical
tomodel
theoretical full-slot winding defined the coefficients k Cu and k . A typical
of PMSM with the designed three-layer winding. The stack is made of M800-50A
ke example of
multilayer windings are the hairpin winding analyzed e.g., in [10].
electrical steel with the Suralac 5012 insulation and the stack length is lFe = 40 mm.
2.2. Physical Model of the Permanent Magnet Synchronous Machines (PMSM) with Three-Layer
Winding
FigureA6.small stator of
Dependency stack with the
coefficients kCudimensions
and kke on theshown in Figure 7 is used for the physical
turns ratio.
model of PMSM with the designed three-layer winding. The stack is made of M800-50A
electrical
2.2. steel
Physical with
Model of the Suralac 5012
the Permanent insulation
Magnet and theMachines
Synchronous stack length is lFe with
(PMSM) = 40 Three-
mm.
Layer The final hand-made stator winding consists of Ncc = 52 and Ncm = 30 turns. The real
Winding
ratio of the number of turns of the central and marginal coil is Ncc /Ncm = 1.73, and the
A small stator stack with the dimensions shown in Figure 7 is used for the physical
relative difference between the ideal value (7) and real value is smaller than 0.1%. The
model of PMSM with the designed three-layer winding. The stack is made of M800-50A
machine is then designed including the calculation of the equivalent circuit parameters [14]
electrical steel with the Suralac 5012 insulation and the stack length is lFe = 40 mm.
and the design is validated by the finite element analysis.
Machines 2021, 9, x FOR PEER REVIEW
234 7 7of
of 15

Figure 7. Dimensions of the stator stack.

The final hand-made stator winding consists of Ncc = 52 and Ncm = 30 turns. The real
ratio of the number of turns of the central and marginal coil is Ncc/Ncm = 1.73̅, and the
relative difference between the ideal value (7) and real value is smaller than 0.1%. The
machine is thenFigure
designed
Figure including
Dimensions
7. Dimensions
7. ofthe
of calculation
the stator
stator stack. of the equivalent circuit parameters
stack.
[14] and the design is validated by the finite element analysis.
The lining of theThe
The lining
slots
final of the slots
is hand-made
achieved is achieved
bystator
the Kapton
winding byconsists
tapethe Kapton
(Figure Ntape
of8a) ccand (Figure
= 52the N8a)
cm = and
andinter-turn the inter-turn
30 turns. The real
insulation
insulation is ensured only
ratio of is
theby ensured only
the double-layer
number by the
of turns ofwire double-layer
the insulation.
central andThewire insulation.
color emphasis
marginal The
coil is Nccof color emphasis
cm = 1.73
/Ndif- of
̅ , and the
different
ferent phases isrelative phases
shown difference is
in Figure 8b shown in
The rotor
between Figure 8b The
yoke value
the ideal rotor
is created yoke
by the
(7) and is created
realconstructionby the construction steel
steelthan 0.1%. The
value is smaller
sheets and
machine is the
sheets and the surface-mounted thensurface-mounted
permanent
designed permanent
magnets
including magnets
arecalculation
the glued are
to theof
yoke,glued
the seeto the yoke,
Figure
equivalent 8c seeparameters
circuit Figure 8c.
[14] and the design is validated by the finite element analysis.
The lining of the slots is achieved by the Kapton tape (Figure 8a) and the inter-turn
insulation is ensured only by the double-layer wire insulation. The color emphasis of dif-
ferent phases is shown in Figure 8b The rotor yoke is created by the construction steel
sheets and the surface-mounted permanent magnets are glued to the yoke, see Figure 8c

(a) (b) (c)

Figure 8. Parts of the

Figure machine:
8. Parts of thestator with partially
machine: inserted
stator with slotinserted
partially insulation layer
slot (a), detailed
insulation layer look at the wound
(a), detailed look stator with
at the wound stator with highlighted coils belonging to different phases (b), and
highlighted coils belonging to different phases (b), and rotor with mounted permanent magnets (c). rotor with mounted
permanent magnets (c).
The machine was built and measured under the no-load and short circuit operation to
The machinemeasure the and
was built (a) circuit
equivalent
measured parameters.
under the no-load Comparison(b) circuit
and short operation (c)of the back
of the measurement
EMF with circuit
to measure the equivalent
Figure 8. Parts
theofFEA results is Comparison
parameters.
the
shown
machine: stator
in Figure
of the
with partially
9ameasurement
Reference
inserted
machine with
of thelayer
slot insulation back double-layer
(a), detailed look
EMF with the FEAwinding
at woundq = stator
the results 1/2 was
withbuilt
is shown in parallel
in Figure
highlighted 9a and the reader
Reference
coils belonging can compare
machine with
to different the and
back-EMF
double-layer
phases (b), waveforms
rotor with mounted
obtained
was builtby
winding q = 1/2 permanent in both windings
parallel
magnets (c).and thein Figurecan
reader 9. compare
The harmonic content ofwaveforms
the back-EMF the back EMF waveforms
obtained by bothobtained
windings by in
theFigure
FEA and measurement
9. The is almost
harmonic content ofidentical
the backforEMFboth windings. Even from the
waveforms
obtained by theshape
FEAThe of the
and waveforms,
measurement
machine was builtit is
is evident
almost
and that the
identical
measured modified
for both
under three layer
thewindings.
no-load winding
Even
and from
short (a) generates
circuit operationa
smaller amount
the shape of thetowaveforms, of spatial
measure theitequivalent
is evident thathigher harmonic
theparameters.
circuit components.
modified three layer winding
Comparison (a) gener-
of the measurement of the back
ates a smaller amount
EMF with of spatial
the FEA higher harmonic
results is showncomponents.
in Figure 9a Reference machine with double-layer
winding q = 1/2 was built in parallel and the reader can compare the back-EMF waveforms
obtained by both windings in Figure 9. The harmonic content of the back EMF waveforms
obtained by the FEA and measurement is almost identical for both windings. Even from
the shape of the waveforms, it is evident that the modified three layer winding (a) gener-
ates a smaller amount of spatial higher harmonic components.
Machines 2021, 9, 234 8 of 15
Machines 2021, 9, x FOR PEER REVIEW 8 of 15

(a) (b)
Figure9.9.Comparison
Figure Comparisonof
ofthe
theback-EMF
back-EMFof
ofthe
thewinding
windingwith
withqq==2/5
2/5 (a)
(a) and
and qq == 1/2
1/2(b).
(b).

3. Thermal
3. Thermal Analysis
Analysis of of the
the Machine
Machine
The thermal
The thermal analysis
analysis has
has been made in two steps. The The non-rotating machine was
loaded by the direct current and measured
loaded by the direct current and measured at at first to obtain the equivalent coefficient of
the thermal conductivity of the handmade winding
the thermal conductivity of the handmade winding placed in the placed in the stator
stator slot.
slot. The
The second
second
test consists
test consists of
of the
the generator
generator operation
operation with
with the
the winding
winding loaded
loaded by
by the
the resistive
resistive load.
load. This
This
thermal analysis
thermal analysis ofof the
the rotating
rotating machine
machineisisfocused
focusedon onthe
thedefinition
definitionofofthe
thereal
realheat
heattransfer
trans-
coefficient
fer of the
coefficient stator
of the andand
stator rotor surface
rotor gap.gap.
surface
3.1. Thermal Analysis of Steady-State Non-Rotating Machine
3.1. Thermal Analysis of Steady-State Non-Rotating Machine
This steady-state thermal analysis of the machine enables to focus only on the losses
This steady-state thermal analysis of the machine enables to focus only on the losses
produced by the stator winding and heat transfer to the stator core neglecting the additional
produced by the stator winding and heat transfer to the stator core neglecting the addi-
cooling effects caused by the rotating rotor. Therefore, this analysis is a proper approach to
tional cooling effects caused by the rotating rotor. Therefore, this analysis is a proper ap-
evaluate the equivalent thermal conductivity of the winding.
proach to evaluate the equivalent thermal conductivity of the winding.
The 3D mathematical model of the machine is created at first for the comparison with
The 3D mathematical model of the machine is created at first for the comparison with
the measurement of the physical model. It is important not to exceed these limits:
the measurement of the physical model. It is important not to exceed these limits:
• maximal temperature of the winding 150 ◦ C defined by the temperature class of used
• maximal temperature of the winding 150 °C defined by the temperature class of used
winding insulation;
• winding insulation;
maximal permanent magnet temperature defined by the Curie temperature of the
• maximal
magnet. permanent magnet temperature defined by the Curie temperature of the
magnet.
The maximal temperature of the winding insulation is the limiting factor in this
The maximal temperature of the winding insulation is the limiting factor in this meas-
measurement.
urement.
The windings of all three phases are connected in series to ensure the uniform dis-
The windings
tribution of all
of the joule three
losses inphases are connected
the machine. in series
The winding to ensure
is loaded bythe uniform
a direct distri-
current to
bution of the joule losses in the machine. The winding is loaded by a direct
suppress the core losses and additional magnetic field-related and frequency-related losses. current to
suppress the core losses
The temperature andpermanent
of the additionalmagnets
magnetic field-related
and especially theandtemperature
frequency-related
of the
losses.
stator core and the winding is measured during the test. The thermographic camera
Theistemperature
picture presented in of the permanent
Figure magnets
10. This method and especially
of temperature the temperature
measurement of the
is chosen for
stator core and the check
a fast temperature winding andiscomparison
measured during the test. dissipation
of the thermal The thermographic camerawith
of the machine pic-
ture is presented
mathematical in Figure 10. This method of temperature measurement is chosen for a
models.
fast temperature
The problem of theand
check comparison
equivalent of the thermal
conduction dissipation
coefficients of reduced
could be the machine
to thewith
2D
mathematical models.
geometry; the temperatures are, therefore, measured from the axial direction respecting the
cross section of the machine. The slot area can be described as a composite material with
different thermal properties of the parts and roughly known position of every single part.
The definition of the equivalent thermal conductivity of the wired coil with relatively small
fill factor of the slot and random positioning of the wires in the slot represents the most
challenging task.
Machines 2021, 9, 234 9 of 15
Machines 2021, 9, x FOR PEER REVIEW 9 of 15

Figure 10. Non-contact measurement of the machine temperature by the thermal camera during the
non-rotating tests loaded by DC (nominal current).

The problem of the equivalent conduction coefficients could be reduced to the 2D

geometry; the temperatures are, therefore, measured from the axial direction respecting
the cross section of the machine. The slot area can be described as a composite material
with different thermal properties of the parts and roughly known position of every single
Figure
part. The 10. Non-contact measurement ofthermal
the machine temperature by the thermal camera during the
Figure 10. definition
Non-contactofmeasurement
the equivalent of the machine conductivity
temperatureof bythethewired
thermal coil withduring
camera relatively the
non-rotating
small fill factortests
of loaded
the slotbyandDC (nominal
random current).
positioning of the wires in the slot represents the
non-rotating tests loaded by DC (nominal current).
most challenging task.
Theequivalent
The
The problem of
equivalent the equivalent
conductivity of theconduction
slot depends
slot depends coefficients
on the
on could be
the location
location reduced
of every
of every wire
wire tointhe
in 2D
the
the
geometry;
slot area.
slot the temperatures
area. It is considered
considered that are,
that the therefore,
the random
random positioningmeasured
positioning of from
of the the
the wires
wires inaxial
in the direction
the slot
slot area respecting
area defines
defines
the cross
many
many sectionof
possibilities
possibilities ofthe
of thearrangement
the machine. The
arrangement ofofslot
every
every area canwire
single
single be
wiredescribed
(touching
(touching as aslot
thethe composite
slot wall,
wall, material
touch-
touching
with
ing other
one different
one other thermal
wire,wire,
etc.). etc.). properties
The winding
The winding of the parts
arrangement
arrangement and roughly known
can be generally
can be generally position
differentdifferent of every
for everyfor single
slotevery
and
part.
slot and
every The definition
every and
machine machine of the equivalent
and itaffects
it directly directly thermal
theaffects
equivalent conductivity
the equivalent of the wired coil
thermal conductivity
thermal conductivity with relatively
of the wholeofslot. the
small slot.
Different
whole fillmodels
factor
Differentof the
of the slot and
location
models ofofrandom
thelocation
the wirespositioning
could of thecould
of thebewires
applied; wiresthebe in applied;
the slot
uniform represents
distribution
the uniform the
of
most
the challenging
wires [26]ofisthe
distribution task.
applied
wiresin this
[26] is paper.
applied in this paper.
The
The
The equivalent conductivity
homogenous
homogenous model
modelofofthe of
theslot the
slot slot
isiscreated;depends
created; thethesloton the
area
slot area islocation
replaced
is replaced ofbyevery
thethe
by wire
concentric in the
concen-
slot
circles
tric area.
circles It is considered
respecting
respecting the area ofthat
the area ofthe
(copper random
and and
(copper wire positioning
insulation
wire insulationof
of the wires
allofconductors in theinslot
all conductors theinarea
the defines
stator slot
stator
many
plus possibilities
Kapton tape of
added the inarrangement
the slot) and of every
the single
surrounding wire
slot plus Kapton tape added in the slot) and the surrounding air [15,27] (see Figure 11). air (touching
[15,27] the
(see slot
Figure wall,
11). touch-
The
ingnumber
number
The oneofother wire, etc.).
conductors
of conductors Thetheir
and their
and winding
area area
of the arrangement
of cross-section
the cross-sectioncan beis generally
is then represented
then different
represented byfor
by circles. every
All
circles.
slot and
dimensions
All dimensions every machine
referrefer
to the to real and it directly
topology;
the real topology; affects
the cross-sectionthe equivalent
the cross-section is shown
is shownthermal
in Figure conductivity
in Figure7. Radius
7. Radius of
of the the
of
whole
equivalent slot.
the equivalent Different
coppercopper models
layerlayerand and of
width the location
of the
width of
of insulation the
the insulation wires
and and could
Kapton Kapton be applied;
tapetape
annular the
annularlayersuniform
are
layers
distribution
defined
are by the
defined byofsum
thesum
the wires
of theof[26]
the is
real applied
materials
real in this
between
materials paper.
the center
between of theofslot
the center theandslotthe
andslot thewall;
slot
these The homogenous
wall; values
these are constant
values model
in both
are constant oftangential
in the slot
both is and
created;
tangential andthe
radial slot area
directions.
radial is replaced by the concen-
directions.
tric circles respecting the area of (copper and wire insulation of all conductors in the stator
slot plus Kapton tape added in the slot) and the surrounding air [15,27] (see Figure 11).
The number of conductors and their area of the cross-section is then represented by circles.
All dimensions refer to the real topology; the cross-section is shown in Figure 7. Radius of
the equivalent
Cu copper layer and width of the insulation and Kapton tape annular layers
are defined Insulation
by the sum of the real materials between the center of the slot and the slot
wall; these values are constant in both tangential and radial directions.
Kapton tape
Air

Cu
Figure 11. Scheme of homogenization of the stator slot with geometrical dimensions.
Figure 11. Scheme of homogenization of the stator slot with geometrical dimensions.
Insulation
The equivalent thermal conductivity of the solid materials in the slot is then calculated
The equivalent thermal conductivity of the solid materials in the slot is then calcu-
as:Kapton tape
lated as: Air b w + bi + b k
λekv = bi bk
, (11)
bw
λw + λ + λ i k

where bw is the equivalent width of the copper layer, bi is the equivalent width of the wire
Figure 11.layer
insulation Scheme ofbhomogenization
and of the
k is the equivalent statorofslot
width thewith geometrical
Kapton dimensions.
tape layer.
This reduction of the stator slot and winding to the concentric circles enables to
obtain The
the equivalent thermal
thermal conductivity
conductivity of the solid
(based materials
on the Equationin the
(11))slot
in is then
both calcu-
radial
lated as:
 Machines 2021, 9, 234 10 of 15

and tangential directions as λekv = 2.48 Wm−1 K−1 . This equivalent thermal conductivity
represents an ideal situation when all wires are currently touching surrounding wires and
the surface of the slot and the heat has to pass from the exact center of the stator slot to its
wall.
In the solved thermal model is on the machine surface and the rotor face set the heat
transfer coefficient to αoutside = 9 Wm−2 K−1 . The other parameters are presented in Table 1.
The ANSYS Workbench is used for the calculation.

Table 1. Used parameters for simulation setup.

Specific Thermal Conductivity-λ Value [W·m−1 ·K−1 ]

Stator core (radial direction) 45
Equivalent conductivity of winding 2.48 (X and Y-axis)
Winding (copper) 380 (Z-axis)
Shaft isotropy thermal conductivity 60
PM isotropy thermal conductivity 7.7
Machines 2021, 9, x FOR PEER REVIEW Insulation thermal conductivity 0.16 2 of 17
Slot insulation-KAPTON tape 0.6

The wires
The second are surrounded
category of increasingby the
theairpower
in thedensity
slot andisthe equivalent
based on theair layerof(Figure
usage new 11)
with in
materials thethe
equivalent
magneticconductivity λair isinsulation,
circuit, winding used as a correction
or permanentfactor betweenetc.
magnets, theMag-
ideal and
netic real situation.
materials withThe air layer
a high levelisofnot directly defined
saturation by the
and small lossreal width[8]ofenable
density the air the
layerre-in the
slot in comparison with the copper and insulation parts; its final
duction of the volume of the stator stack which can be supported by permanent magnets value will be defined
with by the comparison
a high of the measurement
density of energy. and 3D Finite
Properties of insulating Element
materials can Analysis
be improved(FEA).in Both
dif- the
equivalent width of the air layer
ferent ways; insulation systems with higher b a and the equivalent thermal conductivity
temperature class allow the increase of athe λ are the
unknown values. The equivalent width
maximal temperature (up to extreme temperatures ba will
[9]). Insulation systems with high di- (see
be derived from the 3D FEA model
Figure
electric 12) and
strength the the
enable initial
usevalue of the thermal
of a thinner conductivity
insulation is analytically
layer for defined calculated.
input voltage.

Figure 12. FEA simulation results of the obtained equivalent conductivity coefficient.

The maximal operating temperature of the winding which cannot be exceeded during
the measurement is 150 ◦ C. The applied direct current of 1 A generates the joule losses
which excite the steady-state temperatures 64.9 ◦ C for the winding, 41.8 ◦ C for the magnets,
and 44.0 ◦ C for the stator core. The same losses are used in the 3D FEA parametrical
calculation and the final value of the equivalent thermal conductivity of the air λair is
iteratively obtained based on the comparison of the 3D FEA and measurement.
Machines 2021, 9, 234 11 of 15

Comparison of the obtained temperature values is summarized in Table 2. The

temperature distribution for the final value of the λair is shown in Figure 12.

Table 2. Comparison of the calculation and measurement—non-rotating machine.

Measurement Measurement Max. Relative

Temperature [◦ C] FEA
(Contact) (Contactless) Error [%]
Winding 64.9 64.9 61.3 1.2
Permanent magnet 41.9 41.8 40.1 0.5
Stator 43.8 44.0 42.2 3.8
Figure 12. FEA simulation results of the obtained equivalent conductivity coefficient.
The equivalent thermal conductivity of the air of this machine is calculated as
The equivalent
λair = 0.042 Wm−1 K−1thermal conductivity
. This value is used forofspace
the air ofbetween
(air) this machine is and
the coil calculated as slot
the stator λair =
0.042
and Wmyoke
stator
−1 K . and
−1 Thisthe
value is used
maximal for space
relative (air) between
difference (see thethe coil2)and
Table the measurement
of the stator slot and
stator yoke and the maximal relative difference (see the Table 2)
and 3D FEA is smaller than 4%. It is a good result for the non-homogeneous wired coil. of the measurement and
3D FEA is smaller than 4%. It is a good result for the non-homogeneous
The difference between the contact and contactless measurement is the influence of the wired coil. The
difference between the contact and contactless measurement is the influence
emissivity settings of the thermal camera. The two calculated coefficients λekv and λair have of the emis-
asivity
strongsettings
impact of onthe
thethermal
results ofcamera.
the DCThe two calculated
thermal test with a coefficients
stopped machine. λekv and λair have a
strong
Theimpact on the results
final thermal of the DC
conductivity thermal
of the whole testslot
with a stopped
λslot machine.
is calculated by an equation
The final
equivalent to (11)thermal
and theconductivity
final value isofλslot
the=whole
0.54 Wm −1λ
slot −1is
Kslot calculated
. This value fitsbythe
an range
equation
of
equivalent
thermal to (11) andobtained
conductivities the final by
value is λslotapproach
a similar = 0.54 Wm −1K−1. This value fits the range of
presented in [13]. The diagram of
thermal
the conductivities
steps in obtained in
solution is presented byFigure
a similar
13. approach presented in [13]. The diagram of
the steps in solution is presented in Figure 13.

Real model / Dimensions

Calculation of equivalent in both radial and tangential

directions thermal conductivity of the slot

FE analysis / Parametrical calculation of equivalent

thermal conductivity of the air / CFD analysis

Comparison between thermal model and measurement

Results / Solution

Figure 13. Diagram of the solution of the thermal model.

Figure 13. Diagram of the solution of the thermal model.

3.2.Thermal
3.2. ThermalAnalysis
AnalysisofofSteady-State
Steady-StateRotating
RotatingMachine
Machine
Thethermal
The thermalconductivity
conductivityofofthe
theslot
slotrepresents
representsthetheconductive
conductiveheat
heattransfer
transferbetween
between
thewinding
the windingandandthe
thestator
statorcore.
core. The
The cooling
cooling effect
effectofofthe
therotating
rotatingrotor
rotorand
andthermal
thermal
convection between the stator, rotor, and surrounding air are defined by the heat transfer
coefficient α. The precise definition of this coefficient and the boundary conditions influence
the correctness of the results obtained by FEA.
Two elementary equations are used for the exact setting of the heat transfer coefficient.
This can be calculated by the usage of the heat flux density qhtc and temperature difference
as:
qhtc = α( T − Tamb ) , (12)
 convection between the stator, rotor, and surrounding air are defined by the heat transfer
coefficient α. The precise definition of this coefficient and the boundary conditions influ-
ence the correctness of the results obtained by FEA.
Two elementary equations are used for the exact setting of the heat transfer coeffi-
Machines 2021, 9, 234 cient. This can be calculated by the usage of the heat flux density qhtc and temperature
12 of 15
difference as:
𝑞ℎ𝑡𝑐 = α(𝑇 − 𝑇𝑎𝑚𝑏 ) , (12)
whereTT is
where is the temperature of
the temperature ofthe
theanalyzed
analyzedpart partofofthethe machine,
machine, and TambTis
and is ambient
ambambient tem-
temperature. The heat transfer coefficient also defines the thermal heat
perature. The heat transfer coefficient also defines the thermal heat transfer resistance transfer resistance Rp
Ras
p as [28]
[28]  
1 K
Rp = 1 𝐾, (13)
𝑅𝑝 α=· A W[ ] , (13)
𝛼∙𝐴 𝑊
where A is area of the surface.
where TheAcalculation
is area of the surface.
of the heat transfer coefficient α is strongly dependent on many aspects,
The calculation
e.g., the dimensions of the of themachine,
heat transfer coefficient
surface area, andαotheris strongly dependent
mechanical on many
properties, shape, as-
pects, e.g., the dimensions of the machine, surface area, and other
material, speed of the medium, thermal conductivity, etc. The dimensionless quantities aremechanical properties,
shape,
used in material, speedand
the analytical of the medium,
numerical thermal conductivity,
calculations to quantify etc.the The dimensionless
thermal convection; quan-
the
tities are used in the analytical and numerical calculations to quantify
Reynold’s (Re) and Nusselt (Nu) numbers are used in this paper. The formulas used for the thermal convec-
tion; the Reynold’s
estimation of the heat(Re) and Nusselt
transfer (Nu)
coefficient numbers
α are areempirical,
different, used in thisandpaper.
definedThe formulas
differently
used for estimation of the heat transfer coefficient α are different,
in every literature [16,28,29]. Individual empirical relations have to be used for every empirical, and defined
part
differently in every literature [16,28,29]. Individual empirical relations
of the machine. The formula for the Reynold’s number applied to the outer surface of the have to be used for
every part magnet
permanent of the machine.
(rotor of the Themachine)
formula is:
for the Reynold’s number applied to the outer
surface of the permanent magnet (rotor of the machine) is:
D ·v
Re = 𝐷 ∙[−] 𝑣 , (14)
𝑅𝑒 =νk [−] , (14)
𝜈𝑘
whereDDisisthe
where thediameter
diameterofofthe theouter
outerrotor surface,v visisthe
rotorsurface, thesped
spedofofthe
thecoolant
coolantandandνν k kisisthe
the
kinematic viscosity of the air. Nusselt’s number is calculated according
kinematic viscosity of the air. Nusselt’s number is calculated according to [28,29]. to [28,29].
The tested speed of the machine is ns = 1000 rpm. The empirically calculated heat trans-
The tested speed of the machine is − n2s = −1000 rpm. The empirically calculated heat
fer coefficients are αst in = 44 Wm K 1 for the inner stator surface and
transfer coefficients are α st in = 44 Wm−2K−1 for the inner stator surface and αrot out = 24
αrot out = 24 Wm−2 K−1 for the rotor outer surface. These coefficients are based on the
Wm−2K−1 for the rotor outer surface. These coefficients are based on the formulas from
formulas from literature [14,16,17]. These values are confirmed by the 3D computational
literature [14,16,17]. These values are confirmed by the 3D computational fluid dynamics
fluid dynamics (CFD) simulation of the rotating machine (see Figure 14).
(CFD) simulation of the rotating machine (see Figure 14).

Figure14.
Figure 14.Computational
Computationalfluid
fluiddynamics
dynamics(CFD)
(CFD)simulation
simulationresults
resultsofofthe
theheat
heattransfer
transfercoefficient—
coeffi-
cient—rotating machine.
rotating machine.

The calculated equivalent thermal conductivity of the slot and heat transfer coefficients
of the stator and rotor surface are applied to the 3D steady-state FEA of the rotating machine.
The usage of Reynold’s and Nusselt’s numbers respects the laminar or turbulence character
of the air flowing in the air gap. The maximal steady-state output current is defined
to Imax = 1.5A (rms current) at the reference speed 1000 rpm per one phase; this value
and excited losses are used for the analysis. Comparison of the results obtained by the
measurement and 3D FEA is summarized in Table 3 and the temperature distribution is
shown in Figure 15.
Machines 2021, 9, 234 13 of 15

Table 3. Comparison of the calculation and measurement—rotating machine.

Measurement Measurement Max. Relative

Temperature [◦ C] FEA
(Contact) (Contactless) Error [%]
Winding 102.6 105.2 103.2 2.5
Permanent magnet 48.9 48.1 47.2 1.8
Stator 75.2 74.8 76.8 1.0

Figure 15. FEA simulation results of the obtained equivalent conductivity coefficient—rotating
machine.

The heat transfer coefficients were derived in previous calculations and directly used
in this 3D FEA analysis of the rotating machine. The approach used is validated by the
comparison of the temperature distribution obtained by the 3D FEA and measurement of
the loaded rotating machine. The obtained error of measured and calculated values are
smaller than 3%.

4. Conclusions
A physical model of the permanent magnet synchronous machine has a relatively high
number of turns in the slot; the round wire winding is, therefore, used with an optimized
turn ratio of this multilayer winding; The ideal ratio of the number of turns of the central
√
and marginal coil is 3, which corresponds to [20]; the winding factor of spatial harmonic
components ν = 4 km ± 1 is then zero according to Figure 4 and the parasitic components
of the produced magnetomotive force waveform are therefore reduced. The presented
mathematical description of the step curve of the magnetomotive force can be modified
and generally applied to the four-layer hairpin windings. One of the overlooked factors of
these multilayer windings is the influence of the arrangement of the coil sides to the slot
leakage inductance; it has been evaluated for different turn ratios and can be also applied
to the four-layer hairpin winding.
The round-wired coils generally have a small fill factor of the slot and the random
positioning of the wires in the slot. The physical model is used for the evaluation of the
thermal properties of the slot. It is necessary to use the combination of the analytical,
FEA, and CFD calculations, and the results are then iteratively obtained by the usage
of a combination of the presented methods. The obtained thermal conductivity of the
slot is λslot = 0.54 Wm−1 K−1 and the heat transfer coefficients for a rotating machine
Machines 2021, 9, 234 14 of 15

are αst in = 44 Wm−2 K−1 and αrot out = 24 Wm−2 K−1 for inner stator surface and outer
rotor surface, respectively. The maximal stable output power of this physical model is
approximately 30 W at the reference speed 1000 rpm.
The paper presents a method for the analytical description of the magnetic field
produced by a multilayer winding, and an approach to the precise calculation of the slot
leakage inductance, which is necessary to the calculation and prediction of the operation
maps of the machine. The multilayer windings can have different configuration but
presented approach can be modified to different winding.
The calculated value of slot heat transfer coefficient can be used by a reader for a
comparison with their own calculations and measurements of machines with similarly
small slot filling factor. The authors will compare the obtained results with other machines
and the results will be applied to the thermal design and analysis of the machines with
similar winding designed for use in industrial applications.

Author Contributions: Conceptualization, J.L., L.V. and R.P.; methodology, L.V; validation, mea-
surements J.L., L.V.; writing—original draft preparation, J.L., L.V.; writing—review and editing, R.P.;
visualization, J.L.; supervision, R.P. All authors have read and agreed to the published version of the
manuscript.
Funding: This research has been supported by the Ministry of Education, Youth and Sports of the
Czech Republic under the project OP VVV Electrical Engineering Technologies with High-Level of
Embedded Intelligence CZ.02.1.01/0.0/0.0/18_069/0009855 and funding program of the University
of West Bohemia number SGS 2021-021.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Vali, M.; Niknam, T.; Gorginpour, H.; Bahmani-Firouzi, B. Optimal Design Procedure of a High-Torque-Density Dual-Stator
Consequent-Pole Vernier PM Machine. Electr. Eng. 2020, 102, 2637–2653. [CrossRef]
2. Popescu, M.; Foley, I.; Staton, D.A.; Goss, J.E. Multi-Physics Analysis of a High Torque Density Motor for Electric Racing Cars. In
Proceedings of the 2015 IEEE Energy Conversion Congress and Exposition (ECCE), Montreal, QC, Canada, 20–24 September
2015; pp. 6537–6544. [CrossRef]
3. Hruska, K.; Kindl, V.; Pechanek, R. Design of a High-Speed Permanent Magnet Synchronous Motor for Electric Kart. Electr. Eng.
2017, 99, 1141–1150. [CrossRef]
4. Schiestl, M.; Marcolini, F.; Incurvati, M.; Capponi, F.G.; Starz, R.; Caricchi, F.; Rodriguez, A.S.; Wild, L. Development of a High
Power Density Drive System for Unmanned Aerial Vehicles. IEEE Trans. Power Electron. 2021, 36, 3159–3171. [CrossRef]
5. Liu, P.; Feng, S. Integrated Motor and Two-Speed Gearbox Powertrain System Development for Electric Vehicle. In Proceedings
of the 2020 IEEE Energy Conversion Congress and Exposition (ECCE), Detroit, MI, USA, 11–15 October 2020; pp. 1499–1504.
[CrossRef]
6. Ou, J.; Liu, Y.; Breining, P.; Gietzelt, T.; Wunsch, T.; Doppelbauer, M. Experimental Characterization and Feasibility Study on High
Mechanical Strength Electrical Steels for High-Speed Motors Application. IEEE Trans. Ind. Appl. 2021, 57, 284–293. [CrossRef]
7. Wohlers, C.; Juris, P.; Kabelac, S.; Ponick, B. Design and Direct Liquid Cooling of Tooth-Coil Windings. Electr. Eng. 2018, 100,
2299–2308. [CrossRef]
8. Ueno, S.; Enokizono, M.; Mori, Y.; Yamazaki, K. Vector Magnetic Characteristics of Ultra-Thin Electrical Steel Sheet for Develop-
ment of High-Efficiency High-Speed Motor. IEEE Trans. Magn. 2017, 53, 1–4. [CrossRef]
9. Iosif, V.; Roger, D.; Duchesne, S.; Malec, D. An Insulation Solution for Coils of High Temperature Motors (500 ◦ C). In Proceedings
of the 2016 IEEE International Conference on Dielectrics (ICD), Montpellier, France, 3–7 July 2016; Volume 1, pp. 297–300.
[CrossRef]
10. Berardi, G.; Bianchi, N. Design Guideline of an AC Hairpin Winding. In Proceedings of the 2018 XIII International Conference on
Electrical Machines (ICEM), Alexandroupoli, Greece, 3–6 September 2018; pp. 2444–2450. [CrossRef]
11. Wirth, F.; Fleischer, J. Influence of Wire Tolerances on Hairpin Shaping Processes. In Proceedings of the 2019 9th International
Electric Drives Production Conference (EDPC), Esslingen, Germany, 3–4 December 2019. [CrossRef]
12. Jung, D.S.; Kim, Y.H.; Lee, U.H.; Lee, H.D. Optimum Design of the Electric Vehicle Traction Motor Using the Hairpin Winding. In
Proceedings of the 2012 IEEE 75th Vehicular Technology Conference (VTC Spring), Yokohama, Japan, 6–9 May 2012. [CrossRef]
Machines 2021, 9, 234 15 of 15

13. Boglietti, A.; Carpaneto, E.; Cossale, M.; Popescu, M.; Staton, D.; Vaschetto, S. Equivalent Thermal Conductivity Determination of
Winding Insulation System by Fast Experimental Approach. In Proceedings of the 2015 IEEE International Electric Machines &
Drives Conference (IEMDC), Coeur d’Alene, ID, USA, 10–13 May 2016; pp. 1215–1220. [CrossRef]
14. Pyrhonen, J.; Jokinen, T.; Hrabovcova, V. Design of Rotating Electrical Machines, 2nd ed; Wiley: Chichester, UK, 2013; p. 612. ISBN
978-1-118-58157-5.
15. Ayat, S.; Liu, H.; Kulan, M.; Wrobel, R. Estimation of Equivalent Thermal Conductivity for Electrical Windings with High
Conductor Fill Factor. In Proceedings of the 2018 IEEE Energy Conversion Congress and Exposition (ECCE), Portland, OR, USA,
23–27 September 2018; pp. 6529–6536. [CrossRef]
16. Shams Ghahfarokhi, P.; Kallaste, A.; Podgornovs, A.; Belahcen, A.; Vaimann, T.; Asad, B. Determination of Heat Transfer
Coefficient of Finned Housing of a TEFC Variable Speed Motor. Electr. Eng. 2020, 103, 1009–1017. [CrossRef]
17. Shams Ghahfarokhi, P.; Podgornovs, A.; Kallaste, A.; Cardoso, A.J.M.; Belahcen, A.; Vaimann, T.; Asad, B.; Tiismus, H.
Determination of Heat Transfer Coefficient from Housing Surface of a Totally Enclosed Fan-Cooled Machine during Passive
Cooling. Machines 2021, 9, 120. [CrossRef]
18. Bochkarev, I.V.; Bryakin, I.V.; Khramshin, V.R.; Sandybaeva, A.R.; Litsin, K.V. Developing New Thermal Protection Method for
AC Electric Motors. Machines 2021, 9, 51. [CrossRef]
19. Adouni, A.; Marques Cardoso, A.J. Thermal Analysis of Low-Power Three-Phase Induction Motors Operating under Voltage
Unbalance and Inter-Turn Short Circuit Faults. Machines 2021, 9, 2. [CrossRef]
20. Cistelecan, M.V.; Ferreira, F.J.T.E.; Popescu, M. Three Phase Tooth-Concentrated Multiple-Layer Fractional Windings with Low
Space Harmonic Content. In Proceedings of the 2010 IEEE Energy Conversion Congress and Exposition, Atlanta, GA, USA, 12–16
September 2010; pp. 1399–1405. [CrossRef]
21. Jeong, M.-J.; Lee, K.-B.; Pyo, H.-J.; Nam, D.-W.; Kim, W.-H. A Study on the Shape of the Rotor to Improve the Performance of the
Spoke-Type Permanent Magnet Synchronous Motor. Energies 2021, 14, 3758. [CrossRef]
22. Herrmann, P.; Stenzel, P.; Vogele, U.; Endisch, C. Optimization Algorithms for Maximizing the Slot Filling Factor of Technically
Feasible Slot Geometries and Winding Layouts. In Proceedings of the 2016 6th International Electric Drives Production Conference
(EDPC), Nuremberg, Germany, 30 November–1 December 2016; pp. 149–155. [CrossRef]
23. Farshadnia, M.; Dutta, R.; Fletcher, J.E.; Ahsanullah, K.; Rahman, M.F.; Lovatt, H.C. Analysis of MMF and Back-EMF Waveforms
for Fractional-Slot Concentrated-Wound Permanent Magnet Machines. In Proceedings of the 2014 International Conference on
Electrical Machines (ICEM), Berlin, Germany, 2–5 September 2014; pp. 1976–1982. [CrossRef]
24. Heller, B.; Hamata, V. Harmonic Field Effects in Induction Machines; Elsevier: Amsterdam, The Netherlands, 1977; ISBN 044499856X.
25. Vaske, P.; Riggert, J.H. Elektrische Maschinen Und Umformer. Teil 2. Berechnung Elektrischer Maschinen, 7th ed; B.G. Teubner: Stuttgart,
Germany, 1974.
26. Zeaiter, A.; Fenot, M.; Saury, D. Numerical Approach to Determining Windings’ Thermal Conductivity. In Proceedings of the
2018 XIII International Conference on Electrical Machines (ICEM), Alexandroupoli, Greece, 3–6 September 2018; pp. 1291–1296.
[CrossRef]
27. Wrobel, R.; Ayat, S.; Baker, J.L. Analytical Methods for Estimating Equivalent Thermal Conductivity in Impregnated Electrical
Windings Formed Using Litz Wire. In Proceedings of the 2017 IEEE International Electric Machines and Drives Conference
(IEMDC), Miami, FL, USA, 21–24 May 2017. [CrossRef]
28. Incropera, F.P.; DeWitt, D.P.; Bergman, T.L.; Lavine, A.S. Fundamentals of Heat and Mass Transfer, 7th ed.; John Wiley & Sons Inc.:
New York, NY, USA, 2011.
29. Raju, K.S.N. Fluid Mechanics, Heat Transfer, and Mass Transfer: Chemical Engineering Practice; John Wiley & Sons, Inc.: Hoboken, NJ,
USA, 2010; p. 736.