Glenn Research Center
HIGH SPEED PERMANENT MAGNET
SYNCHRONOUS MOTOR / GENERATOR DESIGN
FOR FLYWHEEL APPLICATIONS
Aleksandr Nagorny, Ph.D.National Research Council
Glenn Research Center Outline
• Introduction• Selection of the Rated Point• The major requirements for flywheel M/G in space applications• Meeting design requirements• M/G Main Materials Selection• Lamination Thickness Selection• Modeling tools• M/G Preliminary Design Process• Permanent Magnet Material Selection• The RMxprt Analytical Design Output• Armature Reaction and Demagnetization Calculation• M/G 2D Finite Element Modeling• M/G 3D Finite Element Modeling• Concluding Observations and Recommendations• References• Acknowledgements
Glenn Research Center Introduction
• The motor / generator design is a part of work performed at NASA Glenn Research Center devoted to the development of flywheel modules for use in satellite energy storage and attitude control systems.
• The benefits of flywheel module as an energy storage device in spacecraft application compared to the chemical batteries are the following: higher energy and power densities, deeper depth of discharge, broader operating temperature range.
• Flywheel can be used as a source of momentum for attitude control, giving the opportunity to combine two satellite subsystems into one and reduce the overall volume and mass.
Glenn Research CenterSelection of the Rated Point
• Determination of the output power
ü The output power determination is based on the load profiles during the load cycles.
ü From t=0 to t=60 min it is a charge (motor mode), from t=60 to t=90 min it is a discharge (generator mode).In each point the net torque of M/G is equal
Tnet=TES+TAC
Where TES is the torque component required for the energy storage, and TAC is the component required for the attitude control.
Flywheel Combined Energy Storage and Attitude Control Output Power Determination
-1.5
-1
-0.5
0
0.5
1
0 20 40 60 80 100
Time of Cycle, min
To
rqu
e, N
m
-8000
-6000
-4000
-2000
0
2000
4000
6000
Out
put P
ower
, W
Torque Tes Net Torque Torque ac Output Power
Glenn Research Center
The major requirements for flywheel M/G in space applications
• Relatively high electrical frequency of voltages and currents
• High specific power
• High efficiency, low total losses
• Low THD at the back emf waveform
• Low cogging torque values
• Low rotor losses
• High thermal endurance, ability to operate in vacuum without intensive cooling
Glenn Research Center
Meeting design requirements
High specific power:• Right selection of the M/G configuration• Application of high magnetic energy permanent magnet
materials• Application of high permeability core lamination materials
High efficiency, low total losses:• Choice of an AC permanent magnet synchronous machine
with the zero fundamental frequency rotor magnetic and conductive losses
• Application of core lamination materials with low specific losses
• Application of thin diameter stranded wires for the stator armature conductors to reduce the high frequency skin effect losses
Glenn Research Center Meeting design requirements (cont’d)
Low THD at the back emf waveform and low cogging torque value.
Could be achieved by the reducing the high frequency spatial mmf harmonics by the following methods:
• Make the magnet pole arcs short pitched
• Two layer short pitched stator winding• Skewing of the stator core in axial
direction• Relatively large non-magnetic gap;• Small value for slot opening to slot
pitch ratio• Special shape for the stator teeth
(dummy slots)• Lamination of permanent magnets in
axial direction
Glenn Research Center Meeting design requirements (cont’d)
Low value of the rotor losses:• The main components of the rotor losses :ü Back iron loss;ü Eddy current loss in permanent magnet material;ü M/G carbon fiber sleeve loss;ü Windage loss.
The first three components of the rotor losses are caused by the high frequency spatial mmf harmonics. The measures to reduce them are the same as described on previous slide. The effectivemeasure against the back iron losses is lamination of the back iron core.
Glenn Research Center Meeting design requirements (cont’d)
High thermal endurance:• High thermal endurance can be achieved by using the
appropriate materials for the stator and rotor parts:ü Stator and rotor core lamination materialsü Wire insulationü Slot insulationü Winding parts insulationü Motor leads insulationü Permanent magnet materialsü Rotor carbon fiber composite.
For the current M/G design all the materials except carbon fiber composite have the rated temperature above 200 °C.
Glenn Research CenterM/G Main Materials Selection
Using of prospective materials
The major motor materials that can affect the motor performance are the following
• core ferromagnetic materials• permanent magnet materials• magnet wires• winding insulation
Core ferromagnetic materials
• Two major characteristics of the core ferromagnetic materials can affect the motor performance
ü the maximum saturation flux density
ü specific losses
Glenn Research Center Modeling tools
Ansoft Corporation RMxprt software is used for the preliminary motor design.The advantages of the RMxprt software are the following:
• The ability to get an easy and fast response in a convenient form
• The output data can be easily exported to other Ansoft software (Maxwell 2D, Simplorer)
• The program can perform optimization of the input parameters
Glenn Research Center M/G Preliminary Design Process
The numerous design iterations were completed to meet the motor requirements in motor and generator mode.Different rotor configurations, permanent magnet and core materials and M/G geometry were optimized using the RMxprt parametric analysis mode.
The highest level of the output power in a combination of relatively low back EMF THD level was obtained for the surface mounted arc shaped magnet configuration
Glenn Research Center Permanent Magnet Material Selection
NdFe group has higher remanent magnetization and energy product.
SmCo has a better thermal characteristic.
M/G Output Power in Generator Mode for different PM Materials
5336.96
6075.796387.41
7779.15
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
SmCo 24 SmCo 28 NdFe 30 NdFe 38
Type of Magnet Material
W
Glenn Research Center Lamination Thickness Selection
• At high frequencies (1.2 kHz) and high flux densities (2.0 T) the specific iron loss is proportional to the square of lamination thickness. For the reduction of the iron loss, the lamination thickness should have a low value (in our case 0.004’’).
Specific Iron Loss versus Lamination Thickness for High Saturation Cobalt Iron Alloy at 1200 Hz and 2 T
y = 539416x2 + 5783.6x + 13.285R2 = 1
0
50
100
150
200
250
0 0.0025 0.005 0.0075 0.01 0.0125 0.015
Lamination Thickness, Inch
Spe
cifi
c Lo
ss,
W/k
g
Glenn Research Center The RMxprt Analytical Design Output
6Number of Conductors per Slot:
0.728Length of Stator Core (inch):
2.9Inner Diameter of Stator (inch):
5.75Outer Diameter of Stator (inch):
24Number of Stator Slots:
STATOR DATA
SmCoType of Magnet:
0.27Max. Thickness of Magnet (inch):
0.728Length of Rotor (inch):
1.51Inner Diameter (inch):
0.105Non Magnetic Gap, inch
ROTOR DATA
7.557.55Total Net Weight (lb):
1.4821.452Rated Torque (N.m):
5000050000Synchronous Speed (rpm):
98.16498.168Efficiency (%):
72.01672.594Line Current RMS (A):
0.0150.016Cogging Torque (N.m):
0.9640.964THD of Induced Voltage (%):
115115Operating Temperature (C):
1666.671666.67Frequency (Hz):
44Number of Poles:
6161Rated Voltage (V):
7.67.6Rated Output Power (kW):
GeneratorMode
MotorModeM/G parameters at rated point
Glenn Research Center
Cogging Torque
-0.02
-0.015
-0.01
-0.005
0
0.005
0.01
0.015
0.02
0 50 100 150 200 250 300 350 400
Electric Angle, Degrees
To
rqu
e, N
m
Output Power Versus Torque Angle
-5000
0
5000
10000
15000
20000
25000
30000
0 25 50 75 100 125 150 175 200
Angle, Degrees
Po
wer
, W
Efficiency Versus Torque Angle
0
20
40
60
80
100
0 20 40 60 80 100 120 140 160 180 200
Torque Angle, Degrees
Eff
icie
ncy
, %
Current Versus Torque Angle
0
100
200
300
400
500
600
0 20 40 60 80 100 120 140 160 180 200
Torque Angle, Degrees
Cu
rren
t, A
The RMxprt Analytical Design Output (cont’d)
Glenn Research Center
Flux Density Distribution in the Air Gap (RMxprt)
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
0 50 100 150 200 250 300 350 400
Electrical Angle, Degrees
Flu
x D
ensi
ty, T
The RMxprt Analytical Design Output vs FEA
Flux Density Distribution in the air-gap.0.002" from the surface of mannets FEA
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
0 1 2 3 4
Position in the air-gap, inch
B, T
Flux density distribution in the air-gap, 0.053" from the surface of magnets FEA
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
0 1 2 3 4
Position in the air gap, inch
B, T
Flux density distribution in the air-gap, 0.002" from the stator bore FEA
-1.2
-0.8
-0.4
0
0.4
0.8
1.2
0 1 2 3 4
Position in the air gap, inch
B, T
FEA shows higher flux density in the air-gap than RMxprt
Glenn Research Center
Study of M/G Characteristics During the Duty Cycle
M/G Frequency
0200400600800
10001200140016001800
0 10 20 30 40 50 60 70 80 90 100Time, min
Fre
qu
ency
, Hz
M/G Voltage
0
10
20
30
40
50
60
70
0 10 20 30 40 50 60 70 80 90 100Time, min
Vo
ltag
e, V
M/G Armature Current
-80-70-60-50-40-30-20-10
010203040
0 10 20 30 40 50 60 70 80 90 100
Time,min
Cur
rent
, A
M/G Total Loss
0
20
40
60
80
100
120
0 10 20 30 40 50 60 70 80 90 100Time. min
Lo
ss, W
M/G Output Power
-5000-4000-3000-2000-1000
010002000300040005000
0 10 20 30 40 50 60 70 80 90 100
Time, min
Po
wer
, W
M/G Efficency
93.594.094.595.095.596.096.597.097.598.098.5
0 10 20 30 40 50 60 70 80 90 100Time, min
Eff
icie
ncy,
%
The RMxprt Analytical Design Output (cont’d)
Glenn Research Center
Armature Reaction and Demagnetization Calculation
( )
−+= st
Hcst HcHc ϑϑ
α100
1
Samarium Cobalt (Sintered) S2769 Demagnetization Lines at Different Temperatures
0
0.2
0.4
0.6
0.8
1
1.2
-900000 -800000 -700000 -600000 -500000 -400000 -300000 -200000 -100000 0
Hc, A/m
Br,
T
20°C200°C
• Permanent magnet demagnetization curves are sensitive to the temperature
( )
−+= st
Bst
rBrBr ϑϑα
1001
Glenn Research Center
• The common method to check the demagnetization of the permanent magnets due to the armature reaction is described in T.J. Miller's book [1]. The disadvantage of this method is the assumption that thepermanent magnet pole has uniform saturation.
Φr
Rg
Pm0
Fa
RL
Φm
ΦgΦL
Mrr AB=Φ
m
mrecmo l
AP
µµ0=
gg A
gR
0µ′
=g
m
AA
C =Φrgm
g BRP
CB
)1( += Φ
rgm
grm B
RP
RPB
)1(
1 1
+
+=
rec
mrm
BBH
µµ0
−=−
gm
grrec RP
RPPC
0
11 += µ
m
demrecma l
FB
µµ0=mamload BBB −=
Permanent Magnet Demagnetization Curve hm=.300 inch
0.00
1.10
0.0000
0.9043
0.7512
0 00.10.2
0.30.40.50.6
0.70.80.9
1
1.11.2
-900000 -800000 -700000 -600000 -500000 -400000 -300000 -200000 -100000 0
H,A/m
B,T
Armature Reaction and Demagnetization Calculation (cont’d)
Glenn Research Center
A more accurate way to check the demagnetization is FEA.
• Using Maxwell 2D Transient solver, the solution for the rated load should be determined. The relative rotor-stator position and instantaneous values of the phase currents are determined.
• Then, using position and values obtained by transient solver, the Maxwell 2D Magnetostatic model can be created and the flux density distribution in permanent magnets can be determined.
Armature Reaction and Demagnetization Calculation (cont’d)
Glenn Research CenterArmature Reaction and Demagnetization
Calculation (cont’d)
Flux lines in the motor and flux density vectors without armature reactionIa=0
Glenn Research Center
Armature Reaction and Demagnetization Calculation (cont’d)
Ia=387 A
Due to the armature reaction the level of demagnetization in the corner of the magnet pole is higher than in the other areas of the magnet
Glenn Research CenterArmature Reaction and Demagnetization
Calculation (cont’d)
Ia=581A
Permanent Magnet Flux Density versus Armature Current
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
0 100 200 300 400 500 600 700 800
Phase A Current, A
B, T
Bm av 20°C
B m corn 20°C
The average value of flux density in the magnet is still much more than zero, but the corner area of the magnet is demagnetized
Glenn Research Center M/G 2D Finite Element Modeling
• A more accurate solution of M/G characteristics can be found by using finite element analysis. Thus, the accuracy of the results obtained by RMxprt software could be verified.
• The transient finite element model includes a rotor and stator core, conductors, shaft, magnets, spacers between magnets and a carbon fiber ring. Thus, all components of the rotor losses can be determined with a good accuracy.
Glenn Research Center M/G 2D Finite Element Modeling (cont’d)
• Two different approaches were applied to investigate the M/G characteristics: sinusoidal voltage sources and sinusoidal current sources. Both techniques show similar results for the motor characteristics.
• After transient solution the Magnetostatic solver was used to check the value of the torque, developed for the specified current values.
• Maxwell 2D transient mode solver with the time-stepping approach was employed.
• The effects of saturation, eddy currents, slotting, rotor position and space harmonics are taken into account.
• Because of using two-layer short pitched winding, where two layers of the winding have the angular displacement, there is no mirror symmetry in this machine. Thus there was no opportunity to use the master slave symmetry
Glenn Research Center
M/G 2D Finite Element Modeling (cont’d)
Tav=1.6045 Nm
Glenn Research Center
M/G 2D Finite Element Modeling (cont’d)
∫ ⋅=A
JdAJlPσ1
Eddy current losses were calculated for permanent magnets, carbon fiber ring and spacers between magnets.
Pav=8.1 W
Where σ is the conductivity of material, l is the depth of an eddy current loop in Z direction, A is the surface area, J is a current density.
Could be noticed, that the peaks of eddy current losses are located under slot openings. The value of local eddy current density depends on instantaneous value of the current in the closest slot.
Centers of eddy current losses
Glenn Research Center M/G 3D Finite Element Modeling
The stator slot skewing causes a force in axial direction, that can affect the operation of magnetic bearings. The following technique was employed to determine the value of this force.
1) From Maxwell 2D Transient solver, the relative rotor-stator position and instantaneous values of phase currents are found.
2) Using these position and currents values the Maxwell 3D Magnetostatic model was created and the force values versus phase currents were determined, taking effect of skewing into account.
Glenn Research CenterM/G 3D Finite Element Modeling (cont’d)
Force in Axial Direction Caused by Slot Skewing vs Phase Current
-2
0
2
4
6
8
10
12
0 20 40 60 80 100 120 140
Current, A
Forc
e, N
3D model takes 2 layer winding and slot skewing into account
Glenn Research Center G3 Flywheel M/G Design
Based on the design results presented, M/G as a part of new flywheel G3 module was fully designed and the prototype is planned to be fabricated this year.
Glenn Research Center Concluding Observations and Recommendations
• Ansoft RMxprt, Maxwell 2D and 3D are powerful tools for electrical machine design.
• Some suggestions to make the use of these programs more convenient are given below
1. It would be useful as a part of the RMxprt outputs for the synchronous PM motors and generators to show the demagnetizing curve of the magnets (with the load and no-load lines), the energy product line, and the phasor diagram of the rated point of the machine.
2. The simulation of high frequency synchronous PM machine in transient mode requires relatively small time steps. Because of this it takes a lot of computing time to get steady state results for the transient analysis in 2D and especially in a 3D simulation.
Glenn Research CenterReferences
1. J. R., Jr Hendershot, T. J. E. Miller, “Design of brushless permanent-magnet motors,” Oxford University Press, 1996
Glenn Research Center Acknowledgements
This work was performed while the author held a National Research Council Associateship Award at NASA Glenn Research Center Cleveland Ohio