Content uploaded by Li Li

Author content

All content in this area was uploaded by Li Li on Nov 08, 2017

Content may be subject to copyright.

Abstract — High speed brushless permanent-magnet

generators (HSBPMGs) may be the most suitable choice for small

solar co-generation systems due to a variety of merits. For

instance, they offer substantial reduction in size, and thermally

excellent high-power density, which reduces the running costs with

good performance and reliability. Moreover, high efficiencies i.e.

over 90%, light-weight, low operating temperature, high

insulation, no brushes/slip rings and almost negligible cogging

torque make HSBPMGs ideal for co-generation systems. However,

because of the very high rotor speed and high stator frequency, the

design of HSBPMG is quite different from designing a

conventional generator with low speed and low frequency. As the

speed increases, the losses and temperature go up, and thus careful

attention is needed while selecting the design parameters and

material for the machine. This paper is aimed to use the basic

design process for HSBPMGs running at 60,000rpm, with 6.6kW

capacity keeping the losses minimum by using an appropriate

material and cooling method. Finite element analysis of the

machine is carried by using Motor-CAD simulation software, and

modeling of a prototype machine is presented.

Keywords—high speed brushless permanent magnet generators

(HSBPMG); Co-generation; Motor-CAD.

I.

I

NTRODUCTION

In past few years, increasing demand of green energy and to

minimize the dependence on fossil fuels have provide a route to

worldwide combination of renewable energy generation. Mainly

there are two main technologies which are mostly focused: -

wind and solar. Concentrated solar thermal power (CSP)

integrated with co-generation system is one of the promising

solutions for distributed power generation as discussed in [1]. In

regions with low transmission capabilities smaller power plants

are better idea than to a large power plant due to their cost

effectiveness. In electrical machines theory, the output power is

directly proportional to the speed of rotor, higher speed, would

lead to higher output power with reduced size. Permanent

magnet machines with high rotor speed operations are the best

candidates for small co-generation system as they have simple

structure and higher power density as discussed in [2].

Designing high speed brushless permanent magnet (BPM)

generator is quite different from designing a conventional

generator. In high speed operations, the frequency and the

temperatures of the rotor are very high which leads to higher

losses and need careful attention while selecting the design

parameters as discussed in [3].

Different design topologies of high speed BPM generator

were presented by the researchers in past few years presented in

[4-6]. Different design features of HSBPMG have been

investigated in [7-9]. The potential and limits of HSBPM

machines, selection of magnet, rotor and stator material, rotor

and stator topologies and advantages of using retaining sleeves

are illustrated in [10]. Reference [11] presented the design

solution of high speed permanent magnet (PM) machine for a

flywheel application, also rotor losses reduction techniques are

discussed. A high speed PM machine which operates in elevated

temperature environments with specification of 23kW, 84krpm

with a torque density of 50kNm/m

3

is discussed in [12]. Different

materials for high speed PM machines are discussed in [13]. In

high speed operations, liquid cooling method is more effective

than air cooling, but liquid

in jackets alone cannot effectively

cool the inner rotor and windings, so self-ventilation may be

needed along with liquid cooling. In [14] effective cooling

method is presented and also thermal analysis for different rotor

types are discussed.

The design approach demonstrated in this study is elaborated

for a particular type of high speed PM machine. Basic machine

specifications are set to required power of 660kW, required

speed of 60,000 rpm, and output voltage of 700V. Geometry of

the machine is set to 24 stator slots and 2 poles, and 3-phase

machine. The number of slots per pole per phase is q = 2. There

are number of options that can be set for the rotor topology in

Motor-CAD, U-shaped magnets are used with circular ducts in

this design which help to reduce the high temperature and losses.

Concentrated winding is used to achieve the shortest possible

length of the machine; winding is set to double layer overlapping

and the number of turns per coil is 8 with 224 conductors per

slot. Magnet material used in this design is 6.5%SiFe, which is

excellent to reduce the iron losses and has a very good operating

temperature range. In thermal analysis, water jackets are used in

housing for cooling and windings are set to self-ventilate.

Electromagnetic and thermal analysis of the machine is carried

by using Motor-CAD simulation software. Finally, all the results

are verified by finite element analysis.

II. D

ESIGN PROCEDURE

Designing high speed BPM generator is more

complicated than designing a conventional generator. Because

of high frequency and high losses, all the parameters need to be

selected carefully.

In this paper 6.6kW/60,000rpm HSBPM generator is

designed. In first step, basic requirements of the module need to

be decided, secondly selection of rotor and stator is done. Two

common topologies of stator which are mostly used in high

speed applications: slotted stator, slotless stator. Slotted stators

offer high torque density and high acceleration, selected for the

design. Radial flux interior permanent magnets (IPM) rotor is

selected, as it offers low cogging torque and higher peak-to-

peak ripple torque. Selection of material is very critical in high

speed BPM generator design. A material which offers low

losses and has capability to stand with higher temperature needs

to be chosen. Selection of pole number, number of slots, number

of phases, number of slots per pole per phase and winding

layout are very important constraints which affects the overall

efficiency of designed generator.

Finite element analysis (FEA) is done using Motor-CAD.

Motor-CAD provides the ability to quickly and easily perform

electromagnetic and thermal performance tests on prototype

designs. Results are matched and verified with calculated

values. Finally, thermal analysis is done and temperatures are

calculated. If the results are not verified machine calculated

dimensions, rotor and stator structure and selected material

would be modified.

Design of High Speed Permanent Magnet

Generator for Solar Co-Generation System Using

Motor-CAD

Khurram Shahzad

1*

Youguang Guo

1

Li Li

1

David Dorrell

2

1

Faculty of Engineering & IT, University of Technology Sydney, Australia.

2

University of KwaZulu-Natal, South Africa.

*

Khurram.shahzad@student.uts.edu.au

978-1-5386-3246-8/17/$31.00 © 2017 IEEE

III. PM

ROTOR

DESIGN

Radial flux buried magnets rotor topology is selected for this

design as they are capable in heat removal and appropriate

cooling and high power and speed applicability as discussed in

[15]. The rotor aspect ratio defined as length-to-diameter (L/D)

should not be very low it will cause high stiffness and larger

diameter. To keep the rotor radial size minimum, the optimum

value of L/D is kept 2.51. Rotor tip speed is calculated by rotor

radius and rotational speed; its value lies in 50-200 for PM

machines.

A. calculation of rotor parameters

• Number of poles is critical factor that affects the structure

and performance of the machine. It mainly depends on

the speed, N (rpm) of the machine and electrical

frequency, f (Hz) as shown below.

N⋅ (2p) = 120⋅f

If number of poles is higher, frequency will be high

which causes high losses. Moreover, number of slots per

pole per phase reduces which causes less sinusoidal

waveform. So, number of poles is kept lower i.e. 4 in this

study.

• Air gap flux density depends on magnetic height (h

m

) and

magnetic remnant flux density (B

r

).

Magnet losses can be reduced by using smaller magnets,

i.e. small magnet height. To keep the magnetic field

uniform, magnetic height is greater than the air gap by a

ratio 5-10.

• Number of phases (q) can be determined by the following

equation and it affects machine’s power (P), current (I)

and voltage (V) ratings.

3 phase machines has been selected for this design as it

is the most common form and mostly used in industry

also we get balanced torque in 3-phase machine.

• Once number of slots (N

s

) are known pole pairs (p) and

number of phases (q), slots per pole phase (m) can be

calculated as below.

• To get a sinusoidal voltage waveform and reduced

machine harmonics the number of slots/pole/phase can

be changed.

IV. PM

STATOR

DESIGN

Basically, there are two main stator topologies; slotted and

slot-less stator. Slotted motors are good for high torque density

and high acceleration, while slot-less motors are better option for

smooth operation

and good linearity when operating in a servo

control system. Cogging torque will vary greatly with different

motor designs and steps are typically taken to minimize its

impact, like skewing the magnet or stator laminations. Both

technologies offer large through holes and can be designed for

low profile direct drive applications. Slotted stators are good at

higher acceleration as compared to the slot-less stator which are

not good choice for high speed applications despite some

advantages, slotted stator topology is selected for this design.

Slotting helps to get a narrow air-gap and keeps the main field

winding conductors close to the field magnets to get maximum

flux linkage also provides a limited thermal resistance which

helps to get proper cooling of field windings.

A. Stator windings

Since there are an infinite number of possibilities for pole

and slot count combinations and winding layouts, assumptions

are required to focus or limit the scope so that desirable windings

can be found [16]. The assumptions considered here are machine

has three phases, all slots are filled, each slot has two coil sides

or double layer windings, only balanced windings are considered

all coils have the same number of turns and all span the same

number of slots.

Number of stator slots is 24 (N

s

= 24), number of poles is 4

(N

m

= 4), motor constant is 0.7 (K

m

= 0.7), minimum skew

required to eliminate cogging torque is 1 (α

sk

= 1). Fig 1 shows

the winding layout of the machine.

Fig 1. Winding layout of the machine

Winding type is selected as overlapping; the winding definition

is defined by wire size, which is set to be 0.885mm and 0.800mm

from the metric table, and the material between the liner and

lamination is set as air as shown in picture. Number of strands in

hand is set to 14 and conductor separation is 0.025mm.

Conductor/ slot and conductor/ slot drawn should be same.

V. R

OTOR AND STATOR MATERILA SELECTION

Kind of material selected for rotor and stator of

HSPMG is very critical as it affects the losses and overall

efficiency of the machine. Selection of the material depends on

the application of the machine plus cost, saturation flux,

permeability and core losses. We need the material which offers

minimum losses at high-speed operation to be selected for this

design.

6.5%SiFe is selected as a rotor and stator material, the potential

benefits are reduced iron losses,

Low Magneto-striction, high

permeability, stable quality, non-oriented and operating

temperature range on high speed operation. Table below shows

some features of 6.5%SiFe also known as JNEX core material.

Property value

Saturation flux density 1.6 Tesla.

Typical core loss (50kHz, 100mT) 650mW/cm

3

.

Curie temperature 700°C.

Operating temperature range -30°C to 200°C.

Table 1. Properties of 6.5%SiFe

VI. L

OSSES CALCULATION IN MACHINE

Basic losses in HSPMG can be categorised as follows as

discussed in[17]

1. Stator losses

2. Rotor eddy current losses

3. Winding losses

B

g

=

h

m

h

m

+

g

.B

r

P+jQ =q.V.I

m=

N

s

2.p.q

Stator losses Stator losses made up of copper losses and iron

losses. Copper losses consist of conventional I

2

R loss and stray

losses which are due to the skin effect and proximity effect.

I

2

R

losses can be calculated by the following equation.

=

Stray losses consist of (a) skin effect causing because of the same

source conductors and (b) proximity effect which is because of

the electromagnetic field induced in two adjacent conductors

sharing the one slot.

The intensity of skin effect can be analysed

by skin depth which is a measure of a distance of a current flow

along the surface of the material and reduces in amplitude by a

factor 1/e, where e = 2.71828. Skin depth can be calculated by

the formula given below.

δ=2

ωμ

σ

Losses because of proximity effects of the conductor can be

estimated based on the equation below as given below.

=

(

−1)

Iron losses include hysteresis and eddy current losses,

hysteresis

loss per unit volume in terms of maximum flux density and

frequency can be analysed by the following formula.

=η

The eddy current loss per unit volume at lower frequencies

is given by the following equation.

=

Eddy current loss further divided into two classes (a)

classical eddy current and (b) excess eddy current for more

accurate calculation. So, for a given frequency, the iron losses of

the machine are calculated as below.

=

+

()

+

()

Rotor eddy current Losses

Eddy current losses can be

divided into three categories as below. (a)No-load rotor eddy

current losses caused by the presence of stator slots, (b)On-load

rotor eddy current losses caused by winding’s harmonics (space

harmonics), (c)On-load rotor eddy current loss induced by the

time harmonics of the phase currents (PWM). Generally,

following relationship can be used to calculate the eddy current

losses.

=

/

Since rotor loss caused by time harmonics is dominant in most

applications, increasing the switching frequency and using

external line inductance to reduce current THD is a very

effective way to reduce rotor loss[17].

Winding Losses winding in air cause losses as well, power

is needed to control the drag resistance of rotor cylinder.

Different gases or fluids are also used in between rotor and

stator, so winding losses is a function different operating

conditions such as temperature, density, the pressure of these

fluids/gases and the rotational speed of the shaft. Winding loss

can be estimated by the following equation.

=

There are frictional losses as well because of surface roughness

of rotor and stator should also be taken into account for the more

precise calculation of overall losses.

VII. M

ACHINE

’

S BASIC MODEL AND SIZING

We can model our machine as a phasor circuit as follows. As

the machine is assumed to be balanced so the parameters for one

phase can be determined and applied to other two phases. The

phasor diagram of the machine is shown in figure 2.

Fig 2. Phasor circuit of the machine

Where Ra is the winding’s resistance, Xs is the inductance of

winding, Ea is the back electromotive force (emf) and Va is

terminal voltage of the machine.

Winding resistance is calculated analytically by using the

following equation

.

Flux in the machine is calculated as

By using Faraday’s law we can determine the back emf as given

below.

The air gap inductance is given as

Slot leakage inductance is given as

End turn inductance per phase is given as

Total inductance is the sum of three inductances

Machines are basically sized for rated torque capability

rather than power, output torque is proportional to product of

rotor volume and shear stress, air-gap shear stress is proportional

to the product of air-gap electrical and magnetic loading. Shear

stress is calculated as.

Typically for liquid cooled machines the value of shear stress

is 10-20Psi, we will use 15Psi for our design. We have discussed

the tip speed previously, with the assumed value of 200m/s. we

R

a

=

L

σ

.A

φ

=B

flux

.R

s

.L

st

d

θ

0

π

p

Ea =

Vn.sin np

θ

()

n=1

∞

L

ag

=

λ

i=q

2.4

n

π

.

μ

0

.R

s

.L

st

.N

a

2

.k

wn

2

n

2

.p

2

.g+h

m

()

L

as

=2.p.L

st

.perm 4.N

c

2

m−N

sp

()

+2.N

sp

.Nc

2

self

()

L

slot

=L

as

−2.L

am

.cos 2

π

q

higher odd phases

()

L

e

=

μ

0

.N

c

.N

a

2

.

τ

s

2.In

τ

s

.

π

2.A

s

L

s

=

L

ag

+

L

slot

+

L

e

X

s

=

ω

0

.L

s

τ

∝

K

z

.

B

g

can use the following fundamental power equation to find out

the stack length and rotor radius.

We will assume the L/D ratio to be 2.5, air-gap flux density to

be 0.8T, slot height 15mm and slot fill fraction to be 0.5. We can

calculate all design parameters using MATLAB code and given

in table 2.

The phasor relationship between internal voltages (

), terminal

voltage (

) and the synchronous voltage drop is used to get the

terminal voltage and current of the machine as shown below.

=

−(

.

.)

−

.

.

Once the basic parameters are known in depth analysis is done.

To get a better understanding pre-existing database of the same

kind of machines need to have a look. These data bases are

helpful to choose input parameters. This data can be collected

from several firms including ATE in Germany, Rueland electric

in the USA and E und A in Switzerland given in the appendix,

from all these databases and analysing different machines the

following input parameters are developed.

Table 2. Detailed sizing parameters of the machine

VIII. FINITE

ELEMENT

ANALYSIS

(FEA)

A. Machine’s geometry and windings

Geometry of the machine is set as; housing is round, slot

type parallel tooth in stator. Rotor topology is set to u-shaped

inner magnets with circular rotor ducts for ventilation. Stator

and rotor dimensions in radial view as calculated in matlab. The

resulted geometry is shown in fig 1. Winding type is selected as

overlapping; the winding definition is defined by wire size,

which is set to be 0.885mm and 0.800mm from the metric table

and the material between the liner and lamination is set as air.

Fig 3. Radial and axial view of the machine

MMF harmonics of the machine are analysed, no odd harmonics

are detected in this design which depicts voltage waveform is

symmetric about the centre of the rotor flux. Secondly winding

factor is analysed. Because of the almost sinusoidal counter-

electromotive force (CEMF), the PM machines work very well

with distorted stator magneto-motive force (MMF), where the

high winding factor is of highest priority [18]. In figure 5 it can

be seen that winding factor value is 0.96 which is good to

maximise the torque, avoid unbalanced magnetic pull and to

minimize the rotor losses.

Fig 4. Winding factor of the machine

B. Electromagnetic analysis of the machine

Electromagnetic model is solved and results are analyzed

and compared. Output current and terminal voltage is shown in

the following figures.

The terminal voltage waveform is close to sinusoidal, although

some torque ripples can be observed because of harmonic

contents of line-line voltage and skew effect in air-gap flux

density, it shows a good agreement between analytical and FEA

result.

P

=2.

π

.

R

.

L

ST

.

τ

.V

tip

Machine Size:

Machine Diameter = 0.195 m Machine Length = 0.257 m

Rotor radius = 0.032 m Active length = 0.160 m

Slot Avg Width = 8.050 mm Slot Height = 25.000 mm

Back Iron Thick = 11.200 mm Tooth Width = 8.050 mm

Machine Ratings:

Power Rating = 660.0 kW Speed = 60000 RPM

Va (RMS) = 417 V Current = 527.9 A

Ea (RMS) = 430 V Arm Resistance = 0.00246 ohm

Synch Reactance = 0.194 ohm Synch Induct = 0.015 mH

Stator Cur Den = 1049.3 A/cm2 Tip Speed = 201 m/s

Efficiency = 0.993 Power Factor = 1.000

Phases = 3 Frequency = 2000.0 Hz

Stator Parameters:

Number of Slots = 24 Num Arm Turns = 8

Breadth Factor = 0.966 Pitch Factor = 0.966

Tooth Flux Den = 1.14 T Back Iron = 0.81 T

Slots/pole/phase = 2.00

Rotor Parameters:

Magnet Height = 25.00 mm Magnet Angle = 50.0 degm

Air gap = 4.00 mm Pole Pairs = 2

Magnet Remanence = 1.20 T Air Gap Bg = 0.57 T

Magnet Factor = 1.203 Skew Factor = 0.995

Machine Losses:

Core Loss = 1.5 kW Armature Loss = 2.1 kW

Windage Loss = 1.3 kW Rotor Loss = TBD kW

Machine Weights:

Core =14.06 kg Shaft = 3.96 kg

Magnet = 2.30 kg Armature = 9.98 kg

Services = 4.55 kg Total = 34.85 kg

Fig 5. Output current

Fig 6. Terminal Voltage

Cogging torque harmonic spectrum is analysed, it can be seen

that the dominant harmonics are 48, 96 and then 144 which are

well below 1 % of the rated torque. Although cogging torque is

very low still there is a possibility to reduce cogging torque by

changing width of stator slot opening and to skew stator slot.

Fig 7. Harmonic spectrum of cogging Torque

By solving the e-magnetic model, the flux distribution in rotor

and stator of the machine can be analysed. It is verified that flux

lines are leaving the rotor polar piece nearly perpendicularly

which means that the main field components are radial and

tangential are neglected. Tooth and stator yoke have the flux

density value within the calculated range which is 1.8T. Also

tooth and stator yoke dimensions are good to get the design

objectives

.

Fig 8. Open circuit flux distribution

Open circuit losses are analysed as well, we can verify that the

losses in rotor and stator teeth are well below within the

calculated range. Higher losses can be observed in stator yoke

due to some harmonics present in current going through the

armature winding, another cause of these losses may the skin

effect resulting from the same source conductors.

A 3-phase balanced current is injected in armature winding to

get a sine-wave drive. Phasor diagram in figure 27 is used here

to calculate the required terminal voltage. By knowing the value

of impedance and current voltage can be calculated which has a

sinusoidal waveform which verifies the analytical results.

In on-load flux distribution, it can be verified that machine is

highly saturated, the value of flux density is about 2.2T as can

be seen in the figure below. It confirms that the dimensions of

tooth width and stator yoke are good. Losses are also within the

calculated range.

C. Thermal analysis

Compared with common permanent magnet (PM) machine,

the high-speed PM machine has smaller size and larger power

density. However, owing to high-speed high frequency and

small volume, the PM rotor is more easily to become overheated

so that irreversible demagnetization of the PM is induced.

Traditional cooling methods of machine directly cool the

machine stator; the temperature rise of rotor is reduced through

heat exchange among stator, air gap and rotor. Owing to rapid

temperature rise of high speed PM machine rotor, the method

of indirect cooling rotor does not effectively protect PM rotor

from overheating[19].

Aluminium is selected for housing and end-caps to keep the

overall machine light in weight. Liquid cooling is used as a main

cooling method, water jackets are in housing of the machine to

extract heat from active windings and end windings in the end

space region as well. To directly reduce the temperature,

increase in winding and rotor of the machine self-ventilated

method is used to control the winding temperature.

In thermal analysis, the main cooling options are defined and

the final results are obtained using the lumped thermal circuit at

steady state as shown in fig 12. Duty cycle operation is a

specialized feature, directed to analyse specific exploration

regimes. To perform the necessary sensitivity analyses and

check the important cooling paths is more easy and quick to

look at steady state operation. Then, since the cooling is

defined, the duty cycle analysis can be performed.

The thermal network for Motor-CAD is quite sophisticated with

32 nodes. This is shown in Fig.13; the five nodes below

represent the nodes closest to the measured nodes and were

chosen as being suitable for comparison.

Fig 9. Steady state temperatures

The predicted average winding temperature is 112 °C. The

predicted magnet temperature is 107 °C. These values are below

the initial estimated values of temperatures.

D. Slot FEA

In motor-CAD calibration can be done by the conduction heat

transfer in the stator slot. Figure shows the result with

minimum, average and maximum winding temperature of

105°C, 107°C, and 112°C respectively. We can verify that there

is a relatively good match of temperatures in the slot with

minimum, average and maximum winding temperature shown

in figure, we got after solving the slot FEA. Slightly different

results are observed if we distribute the conductors differently

in the slots. We will get large average and maximum

temperature giving a closer match to the analytical solution as

the conductor touching the slot liner move towards the slot

centre.

Fig 10. Steady state temperature in stator slots

E. Performance charts

. Efficiency map is very useful to get the ideal region of

efficiency where the design is laying. At high speeds operations,

we may get a larger spectrum of efficiency. This is because

copper losses are higher at lower speeds. Therefore, at high

speeds the losses are decreased and efficiency increases, as can

be seen in losses map in figure. As we can see that efficiency of

94% can be achieved in a larger spectrum between the speed of

30,000rpm and 60,000rpm.

Fig 11. efficiency map of the machine

IX.

CONCLUSION

in this paper importance of high speed PM machines in

renewable sector is highlighted. Design strategies and problems

are analyzed. A special design topology is presented to

overcome high cost problem in solar co-generation system. The

HSBPMG running at 60,000rpm having four poles, 24 stator

slots and tooth coil winding is modeled to get comprehensive

and cost-effective manufacturing keeping the electromagnetic

losses as low as possible. 6.5% SiFe is selected as a rotor and

stator material as it offers lowest losses. Losses are kept

minimized by using air-gap length, tooth tip optimization and

wedge

geometry.

FEA is done by using motor-CAD and all the

results show good accord with calculated values. Presented

topology is suitable to use in small co-generation system on

high speed application, this design is able to get design targets,

comprehensive, cost effective and low electromagnetic losses.

R

EFERENCES

[1]

Shah, R., R. Yan, and T.K. Saha. Performance assessment of solar

thermal power plants: a case study in Queensland. in 2014 IEEE

PES General Meeting| Conference & Exposition. 2014. IEEE.

[2]

Fengxiang, W., et al. Design considerations of high-speed PM

generators for micro turbines. in Power System Technology,

2002. Proceedings. PowerCon 2002. International Conference on.

2002. IEEE.

[3]

Cho, H.-W., S.-M. Jang, and S.-K. Choi, A design approach to

reduce rotor losses in high-speed permanent magnet machine for

turbo-compressor. IEEE transactions on magnetics, 2006. 42(10):

p. 3521-3523.

[4]

Arkkio, A., T. Jokinen, and E. Lantto. Induction and permanent-

magnet synchronous machines for high-speed applications. in

2005 International Conference on Electrical Machines and

Systems. 2005. IEEE.

[5]

Kolondzovski, Z., et al., Power limits of high-speed permanent-

magnet electrical machines for compressor applications. IEEE

transactions on Energy Conversion, 2011. 26(1): p. 73-82.

[6]

Arumugam, P., et al., High-Speed Solid Rotor Permanent Magnet

Machines: Concept and Design. IEEE Transactions on

Transportation Electrification, 2016. 2(3): p. 391-400.

[7]

Bianchi, N., S. Bolognani, and F. Luise, Analysis and design of a

PM brushless motor for high-speed operations. IEEE

Transactions on Energy Conversion, 2005. 20(3): p. 629-637.

[8]

Tao, Y., et al. Investigation on structure of stator core and winding

for high speed PM machines. in 2005 International Conference on

Electrical Machines and Systems. 2005. IEEE.

[9]

Wang, F., et al. Design features of high speed PM machines. in

Electrical Machines and Systems, 2003. ICEMS 2003. Sixth

International Conference on. 2003. IEEE.

[10]

Bianchi, N., S. Bolognani, and F. Luise, Potentials and limits of

high-speed PM motors. IEEE Transactions on Industry

Applications, 2004. 40(6): p. 1570-1578.

[11]

Nagorny, A.S., et al. Design aspects of a high speed permanent

magnet synchronous motor/generator for flywheel applications.

in IEEE International Conference on Electric Machines and

Drives, 2005. 2005. IEEE.

[12]

Gerada, D., et al. Design issues of high-speed permanent magnet

machines for high-temperature applications. in Electric Machines

and Drives Conference, 2009. IEMDC'09. IEEE International.

2009. IEEE.

[13]

Uzhegov, N., J. Pyrhonen, and S. Shirinskii. Loss minimization in

high-speed permanent magnet synchronous machines with tooth-

coil windings. in Industrial Electronics Society, IECON 2013-

39th Annual Conference of the IEEE. 2013. IEEE.

[14]

Wiak, S., et al., Determination of critical thermal operations for

high-speed permanent magnet electrical machines. COMPEL-

The international journal for computation and mathematics in

electrical and electronic engineering, 2008. 27(4): p. 720-727.

[15]

Sitapati, K. and R. Krishnan, Performance comparisons of radial

and axial field, permanent-magnet, brushless machines. IEEE

Transactions on Industry Applications, 2001. 37(5): p. 1219-

1226.

[16]

Hanselman, D.C., Brushless permanent magnet motor design.

2003: The Writers' Collective.

[17]

Huynh, C., L. Zheng, and D. Acharya, Losses in high speed

permanent magnet machines used in microturbine applications.

Journal of Engineering for Gas Turbines and Power, 2009. 131(2):

p. 022301.

[18]

Moros, O. and D. Gerling. New flexible harmonic cost effective

concentrated winding topology. in Industrial Electronics Society,

IECON 2015-41st Annual Conference of the IEEE. 2015. IEEE.

[19]

Xing, J.Q., et al. Design and Analysis of Fan-cooling for High

Speed Permanent Magnet Machine Rotor. in Advanced Materials

Research. 2012. Trans Tech Publ