Optimizing Power Density and

Efficiency of a Double-Halbach Array

Permanent-Magnet Ironless
Axial-Flux Motor
Kirsten P. Duffy – University of Toledo / NASA GRC

7/26/2016 Joint Propulsion Conference 1

 Background
Hybrid Electric and Turboelectric Aircraft Propulsion

Boeing SUGAR NASA STARC-ABL

NASA N3X

7/26/2016 Joint Propulsion Conference 2

 Background
Turboelectric Propulsion Benefits
Electric drive = motor + generator + other electrical components

Break-Even on Weight

Each aircraft configuration will

yield combinations of power
density and efficiency required
to achieve net benefit

From Jansen et al. “Turboelectric Aircraft Drive Key Performance Parameters

and Functional Requirements”
7/26/2016 Joint Propulsion Conference 3
 Target Application

• Example – HEIST (Hybrid-Electric Integrated Systems Testbed)

• 31-foot span wing section
• 18 fans directly driven by electric motors
• Motors powered by batteries
• Motor dimensions: 5.5” diameter, 2” length
• Target: 13 kW power at 7200 RPM
Our motor design: target 13 kW/kg and 1% loss

http://climate.nasa.gov/news/2286/leaptech-demonstrates-electric-propulsion-technologies/

7/26/2016 Joint Propulsion Conference 4

 Analysis
Double-Halbach PM Array
Ironless Axial Flux Motor
Upper Halbach Array
Rotor

Windings
Stator

Lower Halbach Array

Rotor
7/26/2016 Joint Propulsion Conference 5
 Analysis
Double-Halbach PM Array
Ironless Axial Flux Motor
Upper Halbach Array

N S N S
S N
N S Windings
S N
N S
S N
N S
S N S N

Lower Halbach Array

7/26/2016 Joint Propulsion Conference 6

 Analysis
Double-Halbach PM Array
Ironless Axial Flux Motor

N S N S
S N
N S
S N
N S
S N
N S
S N S N

Model as 2D Pole Pair

7/26/2016 Joint Propulsion Conference 7

 Analysis
Pole Pair Analysis

xp
xm

ym
yg
yc xc A+ C– B+ A– C+ B–

2D magnetostatic pole pair model allows

for simple equation-based analysis

7/26/2016 Joint Propulsion Conference 8

 Analysis
Pole Pair Analysis
y
xp
xm

ym
By yg
x

sin 𝜖𝜋 𝑛𝑚
𝐵𝑦 = 2𝐵𝑅 𝑒 −𝑘𝑦𝑔 1 − 𝑒 −𝑘𝑦𝑚 cos 𝑘𝑥 cosh 𝑘𝑦
𝜋 𝑛𝑚
𝑥2 𝑦2
𝐹𝑐 = 𝐽Δ𝑟 𝐵𝑦 𝑑𝑥𝑑𝑦
𝑥1 𝑦1 k = 2p/xp
7/26/2016 Joint Propulsion Conference 9
 Analysis
Pole Pair Analysis
Axial Flux in Center of Gap, By (T) 1.5

1.0

0.5

0.0

-0.5
2D FEA
-1.0
Equation
-1.5
0.0 0.2 0.4 0.6 0.8 1.0
Circumferential Distance (x/xp)

7/26/2016 Joint Propulsion Conference 10

 Analysis
Force/Torque/Power

𝑒 −𝑘𝑦𝑔 1 − 𝑒 −𝑘𝑦𝑚 sin 𝜖𝜋 𝑛𝑚 𝑥2 𝑦2

𝐹𝑐 = 2𝐽𝐵𝑅 ∆𝑟𝑦𝑔 𝑦𝑚 sin 𝑘𝑥 sinh 𝑘𝑦
𝑘𝑦𝑔 𝑘𝑦𝑚 𝜋 𝑛𝑚 𝑥1 𝑦1

𝐹𝑝 = 𝐹𝑐 𝑇 = 𝑝𝑟𝑎 𝐹𝑝 𝑃 = 𝑇 𝜔𝑟 = 𝑇 𝑅𝑃𝑀 𝜋 30
𝑐=1

xp
xm

ym
yg
y c xc

k = 2p/xp
7/26/2016 Joint Propulsion Conference 11
 Analysis
Power Density – Based on Magnet Mass
𝑃 𝐽𝐵𝑅 𝑣𝑡𝑖𝑝 −𝑘𝑦 1 − 𝑒 −𝑘𝑦𝑚 sin 𝜖𝜋 𝑛𝑚
∝ 𝑒 𝑔
𝑚𝑚 𝜌𝑚 𝑘𝑦𝑚 𝜋 𝑛𝑚

Small gap / pole size Large gap / pole size

high power density low power density

Ratio of gap to pole size

7/26/2016 Joint Propulsion Conference 12
 Analysis
Power Density – Based on Magnet Mass
𝑃 𝐽𝐵𝑅 𝑣𝑡𝑖𝑝 −𝑘𝑦 1 − 𝑒 −𝑘𝑦𝑚 sin 𝜖𝜋 𝑛𝑚
∝ 𝑒 𝑔
𝑚𝑚 𝜌𝑚 𝑘𝑦𝑚 𝜋 𝑛𝑚

Small magnet thickness

Large magnet thickness
to pole size
to pole size
high power density
low power density

Ratio of magnet thickness to pole size

7/26/2016 Joint Propulsion Conference 13
 Analysis
Power Density – Based on Magnet Mass
𝑃 𝐽𝐵𝑅 𝑣𝑡𝑖𝑝 −𝑘𝑦 1 − 𝑒 −𝑘𝑦𝑚 sin 𝜖𝜋 𝑛𝑚
∝ 𝑒 𝑔
𝑚𝑚 𝜌𝑚 𝑘𝑦𝑚 𝜋 𝑛𝑚
1.2

1.0

e = 1, nm = 2
Sin(p/nm)/(p/nm)

0.8

0.6 e = 1.00
e = 0.75
0.4 e = 0.50 e = 0.5, nm = 2
e = 1.00
0.2 e = 0.75
e = 0.50
0.0
0 5 10 15 20
e = 1, nm = 8
Number of magnets per pole pair, nm
107/26/2016 15 20 Joint Propulsion Conference 14
 Analysis

Parameter
Target power
Target power density
Target loss
Outer diameter
Magnet remanence flux, BR
Current density, J
Electrical frequency, f
Value
13 kW
13 kW/kg
Based on magnet and winding mass only
< 1%
Including magnet and winding losses only
5.5 inches (140 mm)
1.4 T (NdFeB)
3 A/mm2 (natural convection)
to 30 A/mm2 (liquid cooling)
< 2000 Hz
≤ 16 pole pairs at 7200 RPM

7/26/2016 Joint Propulsion Conference 15

 Results
Power
25

20
Motor Power (kW)

15
J = 3 A/mm^2
Low ID/OD
J = 10 A/mm^2
10
J = 20 A/mm^2
J = 30 A/mm^2
5

0
0.2 0.4 0.6 0.8 1.0
Ratio of Motor ID to OD High ID/OD

yc = 3 mm, 16 pole pairs, magnet aspect ratio ym/xm = 1

16 pole pairs  f = 1920 Hz

7/26/2016 Joint Propulsion Conference 16

 Results
Power Density
25

20
Power Density (kW/kg)

15
J = 3 A/mm^2
Low ID/OD
J = 10 A/mm^2
10
J = 20 A/mm^2
J = 30 A/mm^2
5

0
0.2 0.4 0.6 0.8 1.0
High ID/OD
Ratio of Motor ID to OD

yc = 3 mm, 16 pole pairs, magnet aspect ratio ym/xm = 1

16 pole pairs  f = 1920 Hz

7/26/2016 Joint Propulsion Conference 17

 Results 2 𝑉
𝐽𝑟𝑚𝑠 𝑐
I2R Loss Pc 𝑃𝑐 ∝
𝜎𝜂
3.5%

3.0%

2.5%
Resistive Loss (%)

2.0% Low ID/OD

J = 3 A/mm^2
1.5% J = 10 A/mm^2
J = 20 A/mm^2
1.0% J = 30 A/mm^2

0.5%

0.0% High ID/OD

0.2 0.4 0.6 0.8 1.0
Ratio of Motor ID to OD

7/26/2016 Joint Propulsion Conference 18

 Results
Conductor Eddy Loss Pe
2
𝑃𝑒 ∝ 𝜎𝑓 2 𝑑 2 𝐵𝑝𝑘 𝑉𝑐
100.00%
Eddy current loss in conductors

10.00%

1.00% d = 0.50 mm
d = 0.10 mm
d = 0.05 mm
0.10%

0.01%
0.2 0.4 0.6 0.8 1.0
Ratio of Motor ID to OD
7/26/2016 Joint Propulsion Conference 19
 Results
Effect of Magnet Aspect Ratio
25 2.5%

20 2.0%
Power Density (kW/kg)

Resistive Loss
15 1.5%

10 1.0%

Power Density
5 0.5%
Resistive Loss
0 0.0%
0.0 0.5 1.0 1.5 2.0
Ratio of Magnet Axial to Average Circumferential
Length
Rotor ID/OD = 0.6, yc = 3 mm, 16 pole pairs
7/26/2016 Joint Propulsion Conference 20
 Results
Effect of Coil Thickness
16 1.6%

14 1.4%
Power Density (kW/kg)

12 1.2%

Resistive Loss
10 1.0%

8 0.8%

6 0.6%
Power Density
4 0.4%
Resistive Loss
2 0.2%

0 0.0%
0 2 4 6 8 10 12
Coil Thickness (mm)
Rotor ID/OD = 0.6, yc = 3 mm, 16 pole pairs
7/26/2016 Joint Propulsion Conference 21
 Results
Effect of Number of Pole Pairs
13.5

13.0
Power Density (kW/kg)

12.5

12.0

11.5

11.0
0 10 20 30 40
Number of Pole Pairs
B max = 1.0 T
Series1
7/26/2016 Joint Propulsion Conference 22
 Results
Effect of Number of Pole Pairs
2.5%

2.0%

1.5% Small area of

Loss

< 1% loss
1.0%
Eddy current loss for
0.05mm diameter wire
0.5%

0.0%
0 10 20 30 40
Number of Pole Pairs
Resistive Loss Conductor Eddy Loss Resistive + Eddy Loss
Bmax = 1.0 T
7/26/2016 Joint Propulsion Conference 23
 Results
Final Motor Performance
Verified with Maxwell 3D FEA

Parameter Value
Power 13 kW at 7200 RPM
12.8 kW/kg
Power density
Based on magnet and winding mass only
0.85% - conductor resistive loss
Loss 0.11% - conductor eddy current loss
0.02% - magnet eddy current loss (3D FEA)
ID/OD = 0.6, Coil thickness = 3 mm, 16 pole pairs, 20 A/mm2 current
density, and magnet aspect ratio = 1

• Difficult to achieve goal of 13 kW/kg and 1% loss in this configuration

• Required 20 A/mm2 which will require cooling

7/26/2016 Joint Propulsion Conference 24

 Conclusions/Future Work

• Continue to investigate configurations that will

improve efficiency as well as power density
• Design, build and test
• Targets:
• > 1 MW motor
• 13 kW/kg
• 96% efficiency
99% efficiency

7/26/2016 Joint Propulsion Conference 25

 Acknowledgments

This work was funded by NASA:

Advanced Air Vehicle Program
Advanced Air Transport Technology Project
Hybrid Gas-Electric Propulsion Subproject
Amy Jankovsky subproject manager
Thanks to the non-cryogenic motor team members from NASA:
• Yaritza de Jesus-Arce – team leader
• Cheryl Bowman
• Ryan Edwards
• Ralph Jansen
• Peter Kascak
• Andrew Provenza

7/26/2016 Joint Propulsion Conference 26

 Results – Increasing Speed

52 kW Added
weight of
gearbox
212 m/s

26 kW

13 kW
106 m/s
53 m/s

7/26/2016 Joint Propulsion Conference 27

 Results – Increasing Speed
Redesigned for 13 kW with Gearbox
Power Density (kW/kg) 20 2.0%

15 1.5%

Resistive Loss (%)

10 1.0%
212 m/s
53 m/s 106 m/s
5 0.5%

0 0.0%
0 10000 20000 30000 40000
Rotor Speed (RPM)
Power Density Resistive Loss (%)
7/26/2016 Joint Propulsion Conference 28
 3D Transient vs 2D Static Results

14 10%
12 9%
Power Density (kW/kg)

8%
10 7%

Resistive Loss
8 6%
5%
6 4%
4 3%
2%
2 1%
0 0%
Equation-based Equation-based Maxwell 3D Equation-based Equation-based Maxwell 3D
magnetostatic - magnetostatic - transient - magnetostatic - magnetostatic - transient -
optimal design compact coils compact coils optimal design compact coils compact coils

7/26/2016 Joint Propulsion Conference 29

 3D Transient vs 2D Static Results
Eddy Eddy
Current Current
Torque Resistive
Analysis Loss Loss
(N-m) Loss (%)
Conductors Magnets
(%) (%)
Equation-based
magnetostatic 17.3 0.85% 0.11% -
large coils/optimal
Equation-based
magnetostatic 16.3 7.6% 0.06% -
compact coils/high J
Maxwell 3D
magnetostatic 16.6 - - -
compact coils/high J
Maxwell 3D
transient 16.9 8.1% - 0.02%
compact coils/high J

7/26/2016 Joint Propulsion Conference 30