c 2017 Reed Sanchez
DEVELOPMENT OF A HIGH POWER DENSITY ROTOR
BY
REED SANCHEZ
THESIS
Submitted in partial fulfillment of the requirements
for the degree of Master of Science in Electrical and Computer Engineering
in the Graduate College of the
University of Illinois at Urbana-Champaign, 2017
Urbana, Illinois
Adviser:
Associate Professor Kiruba Haran
ABSTRACT
High specific power electric machines are crucial to enabling electric and
hybrid-electric aircraft. In order to meet this need, a high specific power electric machine with an unconventional topology is being developed. The topology creates mechanical challenges including expansion due to high speed, and
thin radial builds that affect rotordynamics. Similarly, the design creates
thermal challenges, of which extracting heat from a small geometry, and the
difficulty of estimating windage losses because of the high speed and large
diameter, are two. These challenges have been addressed through analytical design and a hardware prototype has been validated through testing.
Through this analysis and testing it has been determined that this high specific power electric machine will not fly apart or burn up under the expected
operating conditions.
ii
2 Corinthians 3:18
To my God who provided the ability and will to do this.
iii
ACKNOWLEDGMENTS
First, thank you to Professor Kiruba Haran, who took a chance on a mechanical engineer turning out to be a decent electrical engineer. Thank you
for your leadership in this project, and the opportunity to learn.
Thank you to both the Grainger Center for Electric Machinery and Electromechanics as well as cooperative agreement NNX14AL79A with the NASA
Glenn Research Center for the support. Thank you also to Technical Monitor
Andrew Provenza for your help
I greatly appreciate the help of Test Devices in performing these tests.
This work would not have been possible without you.
Finally, this thesis would be impossible without my family’s support,
specifically that of my soon-to-be wife Sarah. I praise God that you are
in my life.
iv
TABLE OF CONTENTS
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
vi
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
CHAPTER 1 INTRODUCTION . . . . . . . . . . . . . . . . . . . .
1
CHAPTER 2 BACKGROUND INFORMATION
2.1 Specific Power . . . . . . . . . . . . . . . .
2.2 Shear Stress . . . . . . . . . . . . . . . . .
2.3 Mechanical Limits . . . . . . . . . . . . . .
2.4 Electromagnetic Limits . . . . . . . . . . .
.
.
.
.
.
3
3
4
5
6
CHAPTER 3 MOTOR DESIGN SUMMARY . . . . . . . . . . . . .
9
CHAPTER 4 ANALYSIS
4.1 Expansion . . . . .
4.2 Rotordynamics . .
4.3 Cooling . . . . . .
4.4 Windage Losses . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
13
13
14
16
17
CHAPTER 5 TEST SETUP . . . . . . . . . . . . . . . . . . . . . . . 21
CHAPTER 6 TEST RESULTS AND ANALYSIS
6.1 Expansion . . . . . . . . . . . . . . . . . .
6.2 Rotordynamics . . . . . . . . . . . . . . .
6.3 Cooling . . . . . . . . . . . . . . . . . . .
6.4 Windage . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
25
25
26
27
29
CHAPTER 7 CONCLUSION . . . . . . . . . . . . . . . . . . . . . . 32
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
APPENDIX A ENGINEERING DRAWINGS FOR FULL MACHINE 36
A.1 Static Side . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
A.2 Rotating Side . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
v
LIST OF TABLES
2.1
2.2
3.1
Comparison between best in class machines and the motor
presented here for various parameters . . . . . . . . . . . . . .
Typical current densities for electrical machines with different cooling systems [1] . . . . . . . . . . . . . . . . . . . . .
4
7
Key metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
vi
LIST OF FIGURES
3.1
3.2
3.3
3.4
Motor Description . . . . . . . . . .
Manufactured Test Rotor: Internal
Fan Design . . . . . . . . . . . . .
Cooling System Flow . . . . . . . .
.
.
.
.
.
.
.
.
. 9
. 10
. 10
. 11
4.1
4.2
4.3
4.4
4.5
Theoretical Rotor Expansion . . . . . . . . . . . . . . . . .
Rotor Expansion . . . . . . . . . . . . . . . . . . . . . . .
Rotordynamic Campbell Diagram of Test Rotor . . . . . .
Rotor Section View Showing Fan and Exit Holes . . . . . .
Power Lost (Blue) and Flow Rate (Orange) Predicted by
1D Code . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Total Windage Loss Estimation Varying with Speed with
Fan Loss . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
5.1
5.2
5.3
5.4
5.5
Manufactured Test Rotor: External
Test Setup . . . . . . . . . . . . . .
Spin Pit . . . . . . . . . . . . . . .
Expansion Probe Location . . . . .
Test Setup Mounting for Stator . .
.
.
.
.
.
6.1
6.2
6.3
6.4
Expansion Results . . . . . . . . . . . . . . . . . . . . . . .
Rotordynamic Campbell Diagram of Test Rotor and Coupler
Measured Rotordynamic Amplitude vs. Speed . . . . . . . .
Fan Flow Comparison Between Measured Flow at Various
Spin Pit Pressurizations, Desired Flow for Heat Transfer,
and Calculated Flow . . . . . . . . . . . . . . . . . . . . . .
Flow Meter and Air Inlet . . . . . . . . . . . . . . . . . . . .
Measured Deceleration Time from 18,000 rpm at Various
Pressures . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Measured Fan Power and Power Loss at Various Spin Pit
Pressures . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Calculated Windage Loss and Fan Power and Measured
Power Loss at 768 Torr . . . . . . . . . . . . . . . . . . . . .
4.6
6.5
6.6
6.7
6.8
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
14
15
16
17
. . 18
. . 20
.
.
.
.
.
21
22
23
24
24
. 25
. 26
. 27
. 28
. 28
. 29
. 30
. 31
A.1 Stator Assembly Section View . . . . . . . . . . . . . . . . . . 37
vii
A.2 Ground Cylinder Front View Section—Connects to Heat
Sink and Eliminates Torsion, Sets Bearing Distance . . . . . . 38
A.3 Ground Cylinder Top View—Connects to Heat Sink and
and Eliminates Torsion, Sets Bearing Distance . . . . . . . . . 39
A.4 Ground Ring Final Dimensions—Connects to Ground Cylinder, Heat Sink and Fixture to Outside World . . . . . . . . . . 40
A.5 Heat Sink Final Dimensions—Connects to Ground Cylinder and Windings, is Heat Exchanger Between Windings,
Heat Source, and Air . . . . . . . . . . . . . . . . . . . . . . . 41
A.6 Rotor Assembly Section View . . . . . . . . . . . . . . . . . . 43
A.7 Rotor Shell Front Section Final Dimensions from Datum—Described
in Chapter 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
A.8 Rotor Shell Top Final Dimensions—Described in Chapter 3 . . 45
A.9 Rotor Shell Front Air Hole Final Dimensions—described in
Chapter 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
A.10 Fan Front View—Creates Air Flow . . . . . . . . . . . . . . . 47
A.11 Fan Back View—Creates Air Flow . . . . . . . . . . . . . . . 48
A.12 Fan Lid—Eliminates Air Leakage between Fan Channels . . . 49
A.13 Carbon Fiber Ring—Restrains All Parts, Reduces Stress
in All Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
A.14 Lock Nut—Sets Preload on Bearings, Keeps All as One Unit . 51
A.15 Balancing Ring —Restrains Magnets, is Another Surface
to Use for Balancing for Rotordynamics . . . . . . . . . . . . . 52
viii
CHAPTER 1
INTRODUCTION
NASA is funding research for increasing the specific power of electric machines as a key enabling technology for electric aircraft [2]. Current electric
machines, especially non-superconducting machines, do not have the powerto-weight ratio, known as specific power, required for economic 737 class
aircraft flight propulsion. The motor initially presented in [3] is designed to
increase specific power in the megawatt power class to the levels desired by
NASA [2], [4]. The high specific power is achieved through combining a high
tip speed of 264.8 m/s, high electromagnetic loading and a lightweight configuration. All these factors combine to introduce significant mechanical and
thermal challenges. These challenges were analyzed and the design adjusted
to mitigate them. A hardware prototype rotor has been manufactured and
key risks have been retired.
This thesis presents a summary of the machine design as well as the analysis
and validation of rotor expansion, rotordynamics, cooling and air friction or
windage losses. Analytical predictions for key mechanical parameters are
reported in [5] and are refined and validated in this thesis. Much of the
information found in this thesis may also be found in [6].
This thesis begins with background information about high specific power
and high speed electric machines, as well as some of the main challenges
associated with both in chapter 2. Chapter 3 continues with a summary of
the machine design. Chapter 4 consists of machine analysis in four sections.
Section 4.1 presents one challenge, the expansion due to high speed. Another
challenge, the rotordynamics which are complicated due to thin radial builds,
is discussed in section 4.2. Section 4.3 details how forced air cooling was
designed for this high specific power machine. Windage losses, which are
difficult to calculate because of the high speed and large diameter of the
rotor, are considered in section 4.4. Chapter 5 details two test setups which
were used to validate the solutions to expansion, rotordynamic, and cooling
1
challenges, as well as find the windage losses. Test results and an analysis of
those results appear in chapter 6. Finally, chapter 7 concludes the thesis.
2
CHAPTER 2
BACKGROUND INFORMATION
This chapter defines specific power, presents some key metrics, including
shear stress, and gives mechanical and electromagnetic limits on specific
power. A literature review of high specific power electric machines, with
specific power greater than 1 kW/kg, may be found in [4]. All examples
shown here come from that source.
2.1 Specific Power
The specific power of an electric machine is defined as the ratio of output
power to total weight, equation 2.1. This is a key metric for transportation
motors used in trains, automobiles, and airplanes. As the goal of any transportation vehicle is to safely move cargo from one area to another, greater
power means the conveyance can move more. But the greater vehicle weight
means less that it can carry. Thus increasing specific power is crucial to
transportation motor design.
Specif ic P ower =
Output P ower
M achine W eight
(2.1)
The machine presented within this thesis is aimed at achieving the specific
power of 13 kW/kg. For comparison, state of the art for non-cryogenic
specific power electric machines for electric flight is 5.2 kW, shown in [7]. To
achieve this goal, several parameters must be understood, including ν, rotor
tip speed, B, magnetic loading, and K electric loading [4]. These are used
in a sizing equation based on [8] which is used to find output power:
k ∗ K ∗ B ∗ ν ∗ DAG ∗ L
(2.2)
π
In this equation a factor k is included which takes into account machine
P ower =
3
efficiencies and imperfections. DAG is the air gap diameter and L is the
machine length. Note that ν is based on the rotor diameter.
Drotor ω
(2.3)
2
Conventionally ω is the rotational speed in radians per second, and n is the
rotational speed in rpm. The rotor airgap diameters are typically very close,
so the following sizing equation is also valid.
ν=
2
P ower = k ∗ K ∗ B ∗ n ∗ DAG
∗L
(2.4)
Greater understanding of the various parameters introduced above provides insight into why it becomes difficult to increase the specific power. In
Table 2.1 typical numbers for electric machines are cited or specific machines
are used for comparison. While it would be useful to compare these parameters with the current state of the art for specific power in aircraft, 5.2 kW/kg
shown in [7], such information is unavailable. A discussion of the shear stress
is found in section 2.2. The mechanical parameter is presented further in section 2.3. The electromagnetic parameters are further considered in section
2.4.
Table 2.1: Comparison between best in class machines and the motor
presented here for various parameters
Shear Stress
Rotor Tip Speed
Copper Current Density
AG Flux Density
Unit
kPa
m/s
A/mm2
T
Best in Class This Motor
20-35 [9]
23.6
270 [10]
264.8
30 [1]
18
1.0 [11]
0.95
2.2 Shear Stress
The shear stress of a machine is essentially a measure of the per unit electromagnetic force due to electric and magnetic loading. From equation 2.5,
the shear stress is a portion of the output power found in equation 2.2.
Shear Stress = k ∗ K ∗ B
4
(2.5)
Equation 2.6 shows another way to calculate the shear stress, and was used
to compute Table 2.1 for the machine presented in [9]. The range of shear
stresses in Table 2.1 is due to having an unknown rotor radius.
Shear Stress =
T orque
Radius ∗ Rotor surf ace area
(2.6)
This switched reluctance machine found in [9] is rated to 120 hp, or 89.5
kW. It has a magnetic weight of 22 lb, or 10 kg, and a specific power of
9 kW/kg. This specific power shown here is only based on electromagnetic
components, and is less due to other structural parts. With a rated speed of
25,000 rpm, the tip speed is between 114 and 150 m/s. From what we know
of this machine, the Illinois motor outperforms it in most areas other than
potentially the shear stress, especially in terms of the tip speed.
2.3 Mechanical Limits
The main mechanical limit of an electric machine is the tip speed. This limit
is due to two things: the speed of sound and stresses.
As rotor speed approaches the speed of sound the air friction losses grow
beyond what is feasible to mitigate. Thus a hard limit is that of the speed
of sound which is 761 mph, or 340 m/s on a standard day at sea level [12].
This speed decreases significantly with pressure, such as at higher altitudes
This is further discussed in section 4.4.
Maximum rotor stress increases as the square of the speed increases. This
is due to the pseudo-force, commonly called the centrifugal force, which is
due to the rotating reference frame and inertia [13]. This is shown in the
equation below, where Fc is the centrifugal force, m is the mass, and r is the
radius.
mν 2
(2.7)
r
Centrifugal force is related to stress using the definition of stress as a force
per unit area [14], as shown below, where σ is the stress and Ao is the original
cross-sectional area before any deformation:
Fc =
σ=
5
Fc
Ao
(2.8)
This stress is related to strain, the change in length per unit length using
Hooke’s law in equations 2.9 and 2.10, respectively. In the equations, l is the
actual length, lo is the original length, and E is the Young’s modulus or the
elastic modulus.
=
l − lo
lo
(2.9)
σ
(2.10)
Strain based failure is due to increasing tip speed. Shown here, strain
is related to tip speed squared, which makes it a crucial parameter, as it
directly relates a failure mode of the machine with its specific power. Failure
is due to strain as eventually materials can no longer stretch, but ultimate
tensile stress is reported in the literature. Thus while it is strain which causes
failure, it is stress which is used to determine if the material will fail.
There are several ways of dealing with this issue. One is to use a solid
rotor [15], but this limits the design space. Another is to use a retaining
ring, either of a high strength metal or carbon fiber [16]. Similar to other
high specific power machines, the machine presented here uses a high specific
stiffness material—carbon fiber, which reduces the total strain in the radial
direction, thus diminishing stress on all components below it. This allows a
significantly greater tip speed.
The tip speed of 270 m/s, shown in Table 2.1, is the highest cited in the
literature for an electric machine when it was reported in 2013 [10]. The
University of Illinois tip speed is very close to that of [10]. It is interesting
to note though that the rotational speed of [10] was 156,000 rpm, where the
University of Illinois speed is 15,000 rpm. This discrepancy in speed is due
to the difference in rotor radius.
E=
2.4 Electromagnetic Limits
The electromagnetic loading is limited by heat transfer and magnetic saturation. Electric loading, exemplified in copper current density, is limited by
the insulation temperature rating. If the temperature becomes great enough,
the insulation will break down, creating a fault, causing catastrophic failure.
6
Similarly the magnetic loading of a permanent magnet machine is dependent
upon its magnets, and the magnetic saturation of the steel. If permanent
magnets are subjected to high temperatures they become demagnetized, and
thus the rotor becomes unusable. The steel also can only reach a certain
point before becoming ‘saturated’ by the magnetic flux density. Thus, both
electric loading and magnetic loading are limited by temperature and heat
transfer, and magnetic loading is also limited by the saturation.
Temperature reflects the total energy stored in a system, Est , which is
based on the energy which comes into the system, Ein , the energy that exits
the system, Eout , and the energy generated within the system, Eg . This is
exemplified by the following equation [17]:
Est = Ein − Eout + Eg
(2.11)
Electric loading is the energy into the motor system, and the mechanical
output is the main output. The difference is the losses, which are directly
proportional to electric loading. These losses cause an increase in the stored
energy, or temperature, of the machine. This energy has been converted to
heat, and must be removed through cooling of the machine.
There are several types of cooling available. They in include air, oil, aircraft fuel, a water/glycol mixture, or refrigerant cooling [1]. As liquids have
much higher thermal conductivities and specific heat capacities, they are
better coolants than air. This translates to an increased available current
density. Typical current densities for electrical machines with different cooling systems are listed in Table 2.2.
Table 2.2: Typical current densities for electrical machines with different
cooling systems [1]
Cooling System
Totally enclosed machine, natural ventilation
Totally enclosed machine, external blower
Through-cooled machine, external blower
Liquid-cooled machine
Current Density [A/mm2 ]
4.7 to 5.4
7.8 to 10.9
14.0-15.5
23.3 to 30.3
As is noted in [1], an air cooled machine generally has a current density of
between 5 and 15 A/mm2 , which is typical of the low end of current density
of machines. This current density is much less than that available to direct
7
liquid cooling, namely 30 A/mm2 , which is shown in Table 2.1, and in Table
2.2.
The magnetic loading is limited not only by the temperature, but also
by the material flux saturation level. Saturation is essentially the material
reaching a point of diminishing returns. At high magnetic flux intensities, increasing the magnetic field, or H field, results in smaller increases of magnetic
field intensities [18]. This saturation typically means that air-gap flux density for non-cryogenic materials is around 1 T. Air-gap flux densities higher
than this are possible but unlikely.
All of these limits are useful to understand when exploring the machine
design described here.
8
CHAPTER 3
MOTOR DESIGN SUMMARY
The machine design summary consists of three portions. First, the physical
features of the electric machine, on both the rotating and stationary sides, are
described. Second some key machine design choices, used to reduce weight
and increase power, are discussed. Third and finally, some key metrics are
reviewed.
Figure 3.1: Motor Description
Shown in Fig. 3.1, the rotor is made up of a carbon fiber retaining ring,
a titanium shell and permanent magnets. The cantilevered titanium shell is
used to combine the functions of coupling, structural stability and thermal
management while allowing ease of assembly. The shell and carbon fiber ring
together reduce stress in, and retain, the magnets. Magnet stress reduction
is crucial due to the brittle nature of high performance permanent magnets.
Figure 3.2 shows an aluminum ring for balancing operations, and a lock nut
to keep the rotor and stator one unit. The fan, pictured in Figs. 3.2 and
3.3, is used for cooling. It is a centrifugal fan which pulls air through the
9
heat sink and air gap into the center of the fan and then expels that air out
through the holes in the rotor. The air flow path is illustrated in Fig. 3.4.
Figure 3.2: Manufactured Test Rotor: Internal
Figure 3.3: Fan Design
Figure 3.1 also shows the stationary parts. First, the air gap wound high
frequency Litz wire coils provide the rotating magnetic field used by electric
machines to develop shaft power. A high performance ferrite back yoke completes the magnetic circuit, and reduces weight and loss. The 6061 aluminum
heat sink both translates electromagnetic torque to the bearing support, and
provides a path for the cooling flow. The 7075 aluminum bearing support
10
Figure 3.4: Cooling System Flow
fixture supports bearings, translates torque and is the stationary reference.
Hybrid SKF bearings and a wave spring are used for the bearing system
which allows the cantilevered rotor to spin.
Key design choices are made for any high specific power electric machine
to reduce weight and increase power. In this device, weight is reduced by
almost eliminating iron from the machine. Iron reduction is achieved by
going to high frequency using ten pole pairs and high speed. For the stator this allows a thin back yoke, stator teeth to be eliminated and airgap
winding to be employed. On the rotor a Halbach array eliminates iron. The
Halbach array cancels flux on one side and maintains airgap flux density of
approximately 0.9 T in the air gap. The composite ring with a much greater
strength-to-weight ratio continues the weight reduction. In typical configurations, a composite ring increases the magnetic air gap and negatively
impacts electromagnetic performance. This is resolved using an inside-out
architecture, which is rare but does see use in some industrial applications
[19]. High specific power occurs not only through decreasing weight, but also
by maximizing power.
Two factors which affect power are torque and rotational speed. The
torque was optimized through the electro-magnetics presented in [3]. The
rotational and tip speeds are high, as seen in Table 3.1. The tip speed
is limited due to the high centrifugal forces and air friction losses, and is
possible because of the carbon fiber ring. These design choices allow the key
metrics in Table 3.1 to be attained.
11
Table 3.1: Key metrics
Parameters
Rated Power
Rated Efficiency
Specific Power
Total Machine Weight
Machine Active Weight
Insulation Class
Nominal Speed
Tip Speed
Cooling, Forced air
Values
1 MW
96.4%
15 kW/kg
144.2 lbs
75.7 lbs
H
15,000 rpm
264.8 m/s
20 m/s
The above metrics are useful for this program for a number of reasons.
First, and most important, for a 737 class aircraft, a megawatt scale electric
machine is desired. The efficiency, specified to be 96% for the system, is
based on all of the total loss calculations explored in [3]. NASA desires a
specific power of 13 kW/kg [2]. This machine’s specific power is above that
metric. The difference between the active and total machine weight is all
components of the machine which are necessary for the machine to run, but
do not produce power. This includes bearings which allow the machine to
spin, heat sink which conducts excess energy, and carbon fiber ring which
provides structural support. The insulation class is required to handle the
high temperatures which occur through pushing the copper current density
to aerospace levels.
12
CHAPTER 4
ANALYSIS
4.1 Expansion
Expansion caused by centrifugal forces in the rotor as it is torqued to high
rotational speeds can negatively affect electromagnetic performance and rotordynamics, and cause machine failure. As the inside-out rotor is spun, it
expands due to centrifugal forces. This increases the air gap and causes the
equivalent magnetic circuit’s reluctance to rise and B field to decrease; hence,
Lorentz force and torque also decrease. The rotors fundamental modes of vibration, discussed later, can change due to this expansion. From Hooke’s
law, expansion within the rotor is corollary to the stress experienced by the
electric machine. High stress due to large centrifugal forces and thin radial builds corresponds to expansion. This high stress could cause machine
failure.
The expansion is controlled by the following variables: tip speed, ν; average
radial density of components in the rotor, ρ; outer radius of the retaining
ring, Ro ; thickness of the retaining ring, t; Young’s modulus of the carbon
fiber retaining ring, E; and inner radius Ri . The subscripts cf and Ti refer
to the carbon fiber and titanium portions, respectively. These variables are
related in equation 4.1 using a detailed version of Hooke’s law and an integral
expression of the centrifugal force:
∆r =
ν 2 ρeq
Ro2 − Ri2
2(tcf Ecf + tT i ET i )
(4.1)
The expansion and stress have also been modeled extensively through a finite element analysis using ANSYS R . The simulations were set up using
ANSYS R static structural and a ‘fine’ mesh. The main analysis constraints
include the rotational speed applied and rotor bearing supports to the rotor.
13
The results may be seen alongside the measured expansion in Fig. 4.1. The
FEA maximum is the maximum rotor expansion. The FEA at the probe
is what our model assumed would occur at the probe location. Finally, the
analytical is based on equation 4.1.
Figure 4.1: Theoretical Rotor Expansion
These expansion results were included in a combined electromagnetic and
mechanical analysis, to find the effect of rotor expansion on the power produced by the machine. Due to the toothless geometry of the machine, the
magnetic air gap includes not only the physical gap between rotor and stator,
but also the copper windings. Because the magnetic airgap is so large, the
increase of the mechanical air gap has a relatively small effect on the torque
produced, on the order of mT, resulting in a change in torque around 2.5%.
This results in a similarly sized impact on back EMF and reduction in total
power, as seen in Fig. 4.2. The air gap length is in terms of the radius, so
the change is half of that presented in Fig. 4.1.
4.2 Rotordynamics
Rotordynamics, the study of rotating objects, is primarily concerned with
avoiding high amplitude vibrations which can cause catastrophic failure.
These vibrations occur when the synchronous frequency, the frequency at
14
Figure 4.2: Rotor Expansion
which the rotor is spinning, coincides with a natural frequency, a frequency
at which the part tends to vibrate. The thin radial build and high speed of
this machine lead to rotordynamic challenges.
The preliminary design to address the rotordynamics is based on the approach taken in [20], according to which the resonant frequency is controlled
by the following: P , the ratio of Ip , polar moment of inertia to It , transverse
moment of inertia; ω, the synchronous rotational speed; L, the distance
between bearings; and the previously defined It . This formula is useful for
preliminary design and optimization, but must be verified with more complex
formulations.
v
!2
u
P
ω
kL2
Ip ω u
±t
+
ωn = −
2
2
2It
(4.2)
For greater fidelity, the resonant frequencies have been analyzed using
XLRotor, a FEA solver using 1-D beam elements. A map of these frequencies,
and the synchronous frequency and tenth harmonic which corresponds to the
electrical frequency, may be found in Fig. 4.3. The synchronous frequency
intersects with the first natural frequency shown, a backward whirling mode.
The Ip / It ratio is outside the 0.8 to 1.2 range, which indicates that this
mode will not typically be excited. The vibrational mode also occurred well
below the intended operating speed of 15,000 rpm and should not impact the
15
operation speed of the machine. The 10th harmonic, which represents the
electric frequency due to the 20 poles, may have some effect as the machine
speeds up, but at operational speed there should be no effect. At operating
speed there is a margin greater than 10% between synchronous and critical
frequencies, as suggested in [21].
Figure 4.3: Rotordynamic Campbell Diagram of Test Rotor
4.3 Cooling
A key component of the design of an electric machine is the thermal management scheme. To attain high specific power, the electrical and magnetic
loadings of the machine are maximized within thermal constraints. Many
components in this machine are sensitive to an increase in temperature. At
critical temperatures, parts can begin to fail with some potentially catastrophic consequences. Permanent magnets can demagnetize. Coil insulation
can degrade and create shorts. Carbon ring epoxy can soften and cause rotor
structure failure. It is apparent that the generation and extraction of heat
in the machine must be well understood.
While there are several types of cooling, forced air is considered here. It
removes more heat than natural convection and does not require an auxiliary
system like liquid cooling, thus optimizing power for the amount of weight.
16
Part of this forced air system is seen in Fig. 4.4, detailing a section view of
the rotor with its fan and air exit holes. One fillet is used to aid in turning
the air flow from axial to radial. Another fillet between fins is used to relieve
the stress concentration at the fan blade rotor interface.
Figure 4.4: Rotor Section View Showing Fan and Exit Holes
Fan design, shown in Figs. 3.3 and 4.4, comprised several steps. First, a
thermal circuit and ANSYS R model were used to determine the necessary air
speed through the heat sink—20 m/s. Next, computational fluid dynamics
(CFD) was employed to find the pressure drop through the heat sink due to
the air flow—0.6 psi. Using these inputs, the impeller was designed using a
one-dimensional (1D) model [22]. It was overdesigned to account for pressure
drops not included in the CFD model such as those in the end winding
regions. The fan pictured in Fig. 4.4 was designed using the 1D code and
structural analysis. The flow rate calculated by our 1D code is shown in Fig.
4.5.
4.4 Windage Losses
While the extraction of heat has been examined in the cooling, the mechanical
generation of heat must also be considered. Electrical heating is not discussed
in this paper, but some information may be found in [23]. The mechanical
friction losses occur in the ball bearings and at the interface of rotor surfaces
17
Figure 4.5: Power Lost (Blue) and Flow Rate (Orange) Predicted by 1D
Code
and surrounding air. These are the bearing and windage losses, respectively.
Ball bearing losses occur due to frictional movement of the balls rolling in
races through grease. This loss may be calculated through manufacturer
documentation and is not shown here. Windage loss is air friction loss due to
high rotor speed. This loss is especially difficult to calculate as a consequence
of the high speed and cantilevered shape of our machine.
Windage loss is dominated by air friction in two places with the largest
surface area: outer carbon fiber, and air gap. The loss on these surfaces can
be estimated using methods and equations in [24, 25, 26, 27]. In this case
disk windage loss is neglected due to the inner surface actually being the
fan and the outer surface having a proportionally small area. Determining
windage loss begins with finding the Reynolds number, defined here for the
gap between concentric cylinders by equation 4.3.
Re =
ωRδ
ν
(4.3)
In this equation ω is rotating speed, R is rotating surface radius, δ is
gap thickness, and ν is flow dynamic viscous coefficient. Next an associated
friction coefficient, Ccm , may be obtained using example equations 4.4 [24]
and 4.5 [27].
18
Ccm = 0.065(δ/R)0.3 Re−0.2
Ccm =
1
√
−0.8572 + 1.25ln(Re Ccm )
(4.4)
2
(4.5)
Finally using the friction coefficient along with air density, ρ, rotating
surface length L and previously defined variable, the gap windage loss is
found in equation 4.6.
1
(4.6)
P = πρω 3 R4 LCcm
2
All these windage losses result from turbulent flow during high-speed spinning. There is no way to separate the fan power from the windage loss in
testing, so it is included in the windage loss here. It counts as a loss because
while it is necessary for the operation of the machine, it absorbs some electromagnetic power output by the motor. Figure 4.6 presents the estimated
total windage loss varying with rotating speed alongside measured results.
The windage loss prediction boundaries are based on references [24]-[27]. The
speed range is only from 10,000 rpm to 18,000 rpm due to the fact that this
range corresponds to the physical conditions at which these correlations are
valid. These correlations are only valid for turbulent flow, so below a certain
surface speed the correlations will not be valid. Here, uncertainty increases
with rotor surface speed to a point where it is evident that more accurate
models are needed. To achieve a machine efficiency of greater than 96%, the
loss due to surface windage needs to be accurately predicted.
19
Figure 4.6: Total Windage Loss Estimation Varying with Speed with Fan
Loss
20
CHAPTER 5
TEST SETUP
A full size prototype rotor, shown in Fig. 5.1, was built for the purpose of
mitigating risks associated with high speed spinning and for verifying analytical models. As mentioned, these models include mechanical expansion,
rotordynamics, rotating loss, and fan air flow. Because the test was mechanical, not electrical, the permanent magnets were replaced by stainless steel
of a similar density, shown in Fig. 3.2 within the rotor. Two test setups
which used: the first to find expansion, and the second to understand rotordynamics, cooling and windage losses. The first test setup may be seen in
Fig. 5.2.
Figure 5.1: Manufactured Test Rotor: External
Due to the high speed there is a large amount of energy in the rotor when
it is at speed. If this rotor were to break at speed, the result would be
destructive. For this reason these tests occurred in a spin pit, shown in Fig.
21
Figure 5.2: Test Setup
5.3. The rotor is held in the vertical orientation using an arbor. This arbor
connects to the air turbine, which is used to spin the rotor, as pictured in
Fig. 5.2.
The expansion was measured using an eddy current probe from Lion Precision, as shown in Fig. 5.4.
The second test setup used a test stator including bearings, and heat sink.
It was mounted to the test setup with the apparatus shown in Fig. 5.5. This
setup was not usable for the expansion test because eddy current probes were
used to measure the expansion and could not be used at the same time as a
representative heat sink mounted on the test rotor.
The test stator allowed measurement of rotor vibration, flow rate, and
power due to the losses in the motor. The rotor vibration was found through
an eddy probe looking at the spindle shaft. The flow rate was measured
using the flow meter measuring one of the air inlets to the motor. The power
loss was calculated from timing the spin-down of the motor. All of these
measurements will be discussed in the coming sections. Each measurement
occurred at different spin pit pressurizations from vacuum to 768 Torr. These
22
Figure 5.3: Spin Pit
pressure values were chosen to be close to 1, 3/4, 1/2, and 1/4 of an atmosphere. These different pressures were used to better understand the effect
of changing pressure on the measurements.
23
Figure 5.4: Expansion Probe Location
Figure 5.5: Test Setup Mounting for Stator
24
CHAPTER 6
TEST RESULTS AND ANALYSIS
6.1 Expansion
Experimental rotor expansion results are shown in Fig. 6.1. The theoretical and measured values are quite close and will be compared with percent
differences. The FEA maximum expansion has a percent difference of 4.1%,
at the probe the difference is 9% and the analytical formulation was 5.6%
different. All of these are close to the measured expansion. The percent
differences are likely due to not fully capturing the properties of either the
carbon fiber or glue.
Figure 6.1: Expansion Results
Second, a proof test was performed. The rotor was spun up to 18,000 rpm.
This is 20% higher than the intended operating speed. The rotor did not
break, nor did it seem to fail in any way.
25
6.2 Rotordynamics
For the test a separate rotordynamics map was created shown in Fig. 6.2.
This map includes both rotor and coupler as both rotate and need to be
mapped for the test. This map will be compared with the rotordynamic
results.
Figure 6.2: Rotordynamic Campbell Diagram of Test Rotor and Coupler
Rotor vibration was recorded during all phases of the risk mitigation testing. Figure 6.3 shows displacements recorded versus speed for first and last
rotor tests which occurred in a vacuum. There is one clear rotordynamic
mode measured around 8800 rpm, which corresponds to the first forward
whirling mode of Fig. 6.2 at 8,930 rpm with a percent error of around 1%.
Also shown is a steady upward trend in vibration between 12,000 and 16,000
rpm. At this point the reason for increase in vibration is unknown. The
vibration around 15,000 rpm could be due to either the forward tilt whirl
or a first bending mode, and will be determined by further rotordynamic
analysis. This understanding of rotor test physics will refine models of the
final hardware and determine if any fundamental rotor design changes must
be made.
26
Figure 6.3: Measured Rotordynamic Amplitude vs. Speed
6.3 Cooling
The cooling flow was also measured. Cooling flows as a function of rotor
speed for various spin chamber pressurization levels are shown alongside an
analytical prediction in Fig. 6.4. Clearly the measured flow is not as much
as predicted. Cooling flow is crucial in this electric machine design as a lack
of flow will restrict the amount of electrical current which can be sent to
the stator coils, thus limiting the power output of the device. While this
measurement is discouraging, all hope is not lost. Differences between test
environment and anticipated use environment reduce the relevance of this
measurement for the future use of this configuration.
Several details of the test design may have restricted flow though the machine and reduced the relevance of the flow measurement. First, the measurement is suspect, as the chosen flow sensor should be placed within a pipe
rather than in a constriction. Second, the test setup had significantly increased pressure drops compared to the intended application, which decrease
the flow rate. This occurs at the inlet, where the total area of the four holes
is less than heat sink available area, seen in Fig. 6.5. A further pressure
increase could occur at the exit when flow goes from rotating to exiting radially. Third, the pressure ratio is opposite of how compressors are designed to
work. The inlet area of a compressor is designed to be greater than the outlet
27
Figure 6.4: Fan Flow Comparison Between Measured Flow at Various Spin
Pit Pressurizations, Desired Flow for Heat Transfer, and Calculated Flow
area. This difference causes the compressor to work. In contrast, the test
setup inlet is smaller than the test setup outlet. Thus the measurement itself
being suspect, the increase in pressure drops, and the disagreement between
compressor design and test setup all contribute to an under-performing fan
flow.
Figure 6.5: Flow Meter and Air Inlet
The testing environment was not ideal for the flow measurement, but was
required for other measurements and safety. First, in order to keep structural
modes from influencing rotordynamic modes, the stationary portion of the
test setup had to be stiff. This stiffness required significant material which
28
blocked air flow. This was also useful from a safety standpoint to protect
against burst. Thus, while the test did not allow accurate flow measurements
for the use environment, it gives confidence in the safety of the part, so future
flow measurements can occur.
6.4 Windage
Windage losses cannot be measured directly but can be determined from rotor deceleration data. Rotor power can be calculated as in a rotational power
equation 6.1 [28]. Finding the change in speed vs. time was accomplished
by bringing the setup to 18,000 rpm and stopping the power going into the
drive. This is shown in Fig. 6.6, which illustrates the time it takes for the
rotor to spin down from 18,000 rpm at different pressures.
Figure 6.6: Measured Deceleration Time from 18,000 rpm at Various
Pressures
During this time the only power acting on the rotor was mechanical friction
losses, either due to the bearings or the windage losses. This occurred for
several pressure levels, including at vacuum. The windage loss was then
found by subtracting the power from the vacuum trial from the other trials,
essentially taking out all other system losses. The calculated windage loss
and fan power at various pressures are seen in Fig. 6.7.
29
P = −ω ∗ J
dω
dt
(6.1)
Figure 6.7: Measured Fan Power and Power Loss at Various Spin Pit
Pressures
The windage loss and fan power were also compared with the expected
values. As is clear, the windage loss and fan are within the error bounds. It
must be noted that the fan power may be less than anticipated during full
testing, but is well within the error bounds as seen in Fig. 6.8.
30
Figure 6.8: Calculated Windage Loss and Fan Power and Measured Power
Loss at 768 Torr
31
CHAPTER 7
CONCLUSION
The main goal of this test, to reduce risk, has been accomplished. A main
risk, caused by the high tip speed atypical in industry, is found in the high
internal stresses and structural integrity issues associated with the expansion.
This risk has been mitigated through expansion measurement and a proof test
to 20% over speed. Another risk, that the rotor would shake itself apart due
to rotordynamics, was reduced in this test. Risk that power loss due to air
friction may be greater than anticipated has also been reduced. In addition,
this test verified most analytical models presented here. These results will
be used to create more accurate motor performance models, and influence
some redesign. Finally, this test highlights a need for further rotordynamic
analysis and flow testing, and lays the groundwork for future full machine
testing.
32
REFERENCES
[1] J. Gieras, “New applications of synchronous generators,” Przeglad Elektrotechniczny (Electrical Review), vol. 88, 2012.
[2] R. D. Rosario, “A future with hybrid electric propulsion systems: A
NASA perspective,” Turbine Engine Technology Symposium Strategic
Visions Workshop, 2014.
[3] A. Yoon, X. Yi, J. Martin, Y. Chen, and K. Haran, “A high-speed, highfrequency, air-core PM machine for aircraft application,” IEEE Power
and Energy Conference at Illinois (PECI), 2016.
[4] X. Zhang, “High-specific-power electric machines for electrified transportation applications technology options,” IEEE ECCE, 2016.
[5] Y. Chen, R. Sanchez, A. Yoon, and K. Haran, “Mechanical design considerations of an iron-less, high specific power electric machine,” IEEE
Transactions on Transportation Electrification, 2017, accepted for publication but not fully edited.
[6] R. Sanchez, A. Yoon, X. Yi, L. Zheng, Y. Chen, K. Haran, A. Provenza,
and J. Veres, “Mechanical validation of high power density external
cantilevered rotor,” Industrial Application Society Transactions, to be
published.
[7] Siemens, “World-record electric motor for aircraft,” January 2017.
[8] T. Lipo, Introduction to AC Machine Design.
tronics Research Center, 2015.
Wisconsin Power Elec-
[9] A. Radun, “High power density switched reluctance motor drive for
aerospace applications,” IEEE Industry Applications Society Annual
Meeting, pp. 568–573, 1989.
[10] A. Borisavljevic, “Limits, modeling and design of high-speed permanent
magnet machines,” Delft University of Technology, 2013.
33
[11] G.
Long,
“High
efficiency,
high
power
density
electric
motors,”
LaunchPoint
Technologies.
[Online]. Available: https://cafe.foundation/v2/pdf tech/MPG.engines/
HE HP electric motors Long 20090929.pdf
[12] NASA, “Speed of sound,” November 2017. [Online]. Available:
https://www.grc.nasa.gov/www/k-12/airplane/sound.html
[13] “Centrifugal force,” 2017, University of Virginia Physics Show. [Online].
Available: http://phun.physics.virginia.edu/topics/centrifugal.html
[14] R. L. Norton, Machine Design, An Integrated Approach, 2nd ed.
Prentice-Hall, 2000.
[15] J. F. Gieras and J. Saari, “Rotor integrity design for a high-speed modular air-cored axial-flux permanent-magnet generator,” IEEE Transactions on Industrial Electronics, vol. 59, no. 6, pp. 2689–2700, June 2012.
[16] W. Fei, P. C. K. Luk, and T. S. El-Hasan, “Performance calculation
for a high-speed solid-rotor induction motor,” IEEE Transactions on
Industrial Electronics, vol. 58, no. 9, pp. 3848–3858, September 2011.
[17] T. L. Bergman, A. S. Lavine, F. P. Incropera, and D. P. Dewitt, Fundamentals of Heat and Mass Transfer, 7th ed. John Wiley and Sons,
Inc, 2011.
[18] Dura Magnetics, Inc., “Magnetic saturation: Understanding practical
limitations to how much induced magnetism can be achieved in a workpiece,” October 2015.
[19] K. R. Weeber, M. R. Shah, K. Sivasubramaniam, A. El-Refaie, R. Qu,
C. Stephens, and S. Galioto, “Advanced permanent magnet machines
for a wide range of industrial applications,” Power and Energy Society
General Meeting, 2010 IEEE, 2010.
[20] S. Y. Yoon, Z. Lin, and P. E. Allaire, Control of Surge in Centrifugal
Compressors by Active Magnetic Bearings. Springer London, 2013.
[21] W. Tong, Mechanical Design of Electric Motors. CRC Press, 2014.
[22] J. P. Veres, “Axial and centrifugal compressor mean line flow analysis
method,” 47th AIAA Aerospace Sciences Meeting Including The New
Horizons Forum and Aerospace Exposition, 2009.
[23] J. Martin, A. Yoon, A. Jin, and K. Haran, “High-frequency litz air-gap
windings for high-power density electrical machines,” Electric Power
Components and Systems, pp. 798–805, 2017.
34
[24] E. Bilgen and R. Boulos, “Functional dependence of torque coefficient of
coaxial cylinders on gap width and Reynolds numbers,” JSME, vol. 95,
no. 1, pp. 122–126, 1973.
[25] P. Childs, Rotating Flow. Butterworth-Heinemann, 2010.
[26] E. Graf et al., “Segregation of windage and core losses for high speed
/ frequency, permanent magnet machines,” 2008. [Online]. Available:
http://navalengineers.net/Proceedings/EMTS2008/Papers/Graf.pdf
[27] J. Vrancik, “Prediction of windage power loss in alternators,” NASA
Technical Note, Tech. Rep., 1968.
[28] S. D. Umans, Fitzgerald & Kingsley’s Electric Machinery. McGraw-Hill
Education, 2013.
35
APPENDIX A
ENGINEERING DRAWINGS FOR FULL
MACHINE
A.1 Static Side
Pictured in the following drawings is all that is needed to create the full
machine . This is split up into both static and rotating sides. The static side
begins in Fig. A.1, showing the stator when it is fully assembled. Figures A.2
and A.3 show the ground cylinder. Figure A.4 shows the final dimensions of
the ground ring. Finally, Fig. A.5 shows the heat sink final dimensions.
36
Figure A.1: Stator Assembly Section View
37
1X PRELOAD SPRING
1xYOKE
19x HEAT SINK SECTIONS
ASSEMBLY FEATURES:
HEAT SINK SECTIONS TO GROUND RING - BOLTED ALL THREAD
HEAT SINK AND GROUND RING TO GROUND CYLINDER - 5 mil SHRINK FIT
YOKE TO HEAT SINK - SHRINK FIT
WIRE TO YOKE GLUE
-A-
TOLERANCE
APPROVED
CHECKED .00 .01
.000 .003
DRAWN
R.S.
5/11/2017
SCALE
A3
SIZE
.5
DWG NO.
TITLE
STATOR ASSEMBLY
1 OF 1
UNIVERSITY OF ILLINOIS
1x GROUND CYLINDER
1x GROUND RING
30x WINDINGS
REV
Figure A.2: Ground Cylinder Front View Section—Connects to Heat Sink
and Eliminates Torsion, Sets Bearing Distance
38
-A-
11.560
10.115
10.615
3.440
4.585
Ø5.712
Ø5.118
Ø5.512
Ø6.112
Ø5.512
Ø5.118
R.100
6.030
TOLERANCE
APPROVED
CHECKED .00 .01
.000 .003
DRAWN
R.S.
5/10/2017
DWG NO.
.75
A3
SIZE
SCALE
TITLE
GROUND CYLINDER
FRONT SECTION DIMENTIONS
1 OF 2
UNIVERSITY OF ILLINOIS
REFERENCE
DIMENSION
NEED TO CONTROL FOR BEARINGS
0.001
MAKE FROM AL7075
REV
Figure A.3: Ground Cylinder Top View—Connects to Heat Sink and and
Eliminates Torsion, Sets Bearing Distance
39
Ø6.312
Ø5.118
4x1.50
3.066
TOLERANCE
APPROVED
CHECKED .00 .01
.000 .003
DRAWN
R.S.
5/10/2017
DWG NO.
1
A3
SIZE
SCALE
TITLE
GROUND CYLINDER
TOP FINAL DIMENTIONS
2 OF 2
UNIVERSITY OF ILLINOIS
MAKE FROM AL 7075
SHRINK FIT 5 mil
REV
Figure A.4: Ground Ring Final Dimensions—Connects to Ground Cylinder,
Heat Sink and Fixture to Outside World
40
8x
3.45
3/8x24 THREADS
22.5°
4x
1/16
4x
1.50
Ø7.500
THREADED
MOTOR FIXTURE FOR TEST
1/4x20 THREADS
TOLERANCE
APPROVED
CHECKED .00 .01
.000 .003
DRAWN
R.S.
5/9/2017
1.00
DWG NO.
.75
A3
SIZE
SCALE
TITLE
GROUND RING
FINAL DIMENSIONS
3 OF 3
UNIVERSITY OF ILLINOIS
MAKE FROM 7075
GROUND PLATE
REV
Figure A.5: Heat Sink Final Dimensions—Connects to Ground Cylinder
and Windings, is Heat Exchanger Between Windings, Heat Source, and Air
41
3.180
3.316
9.975
.15
SCALE 0.750
.30
4x
5.4137° REF
.0625 FOR PINS
4x .25
FOR ALL THREAD
R.25
6.131
1.500
R3.16
.50
TOLERANCE
APPROVED
CHECKED .00 .01
.000 .003
DRAWN
R.S.
5/9/2017
SCALE
A3
SIZE
TITLE
HEAT SINK
FINAL DIMS
.75
DWG NO.
3 OF 3
UNIVERSITY OF ILLINOIS
MAKE FROM AL6061
GROUND PLATE
REV
A.2 Rotating Side
Figures A.6 to A.15 detail the design drawings for the rotor. Here, Fig. A.6
details the full rotor. Figures A.7 through A.9 present all final dimensions for
the rotor titanium shell. The fan is shown in Figs. A.10 and A.11 alongside
the fan lid in A.12. Finally, the carbon fiber ring, lock nut, and balancing
ring are detailed in Figs. A.13, A.14, and A.15 respectively.
42
Figure A.6: Rotor Assembly Section View
43
SECTION C-C
SCALE 0.250
ASSEMBLY FEATURES
SHELL TO CARBON FIBER - SHRINK FIT 8 MILS
FAN LID TO FAN - PEAN ENDS OF PINS IN FAN
FAN TO SHELL - ALIGN BOSS PINS AND BOLT
MAGNETS TO SHELL - SEND TO ARNOLD
BALANCING RING TO SHELL - SHRINK FIT - 8 MIL
TOLERANCE
APPROVED
CHECKED .00 .01
.000 .003
DRAWN
R.S.
5/12/2017
1X LOCK NUT
DWG NO.
.5
A3
SIZE
SCALE
TITLE
ROTOR ASSEMBLY
1/2 SECTION
1 OF 1
UNIVERSITY OF ILLINOIS
1xCARBON FIBER RING
REV
120 PIECES X 3 LAYERS MAGNETS
1xFAN LID
1xFAN
1X BALANCING RING
-B-
1xSHELL
Figure A.7: Rotor Shell Front Section Final Dimensions from
Datum—Described in Chapter 3
44
3.357 MAJOR DIAMETER
.5 in ENGEGEMENT
LEFT HAND THREAD
2.235
2.835
3.780
SEE DETAIL A
REFERENCE
DIMENTIONS
.075
R.250
R.250
.046
9.915
9.800
10.860
10.160
11.160
R3/32
.005
R.250
R.250
.218 .246
DETAIL A
SCALE 2.000
A
2x
12.530
12.130
11.730
3.823
3.543
4.500
2.750
0.0002
3.300
.700
.400
B
APPROVED
CHECKED
DRAWN
R.S.
4/24/2017
SCALE
A3
SIZE
0.5
DWG NO.
4 OF 5
TITLE
ROTOR SHELL FRONT SECTION
FINAL DIMENTIONS FROM DATUM
UNIVERSITY OF ILLINOIS
REV
SECTION A1-A1
SCALE 0.500
11.560
9.325
TOLERANCE
0.00 = 0.01
0.001 = 0.003
0.001
UNLESS OTHERWISE
SPECIFIED
8.725
0.0005 A
7.780
1.645 1.760
BEARING
B
SURFACE
1.400
10 mil RELIEF
CF RING NOT SHOWN HERE
Figure A.8: Rotor Shell Top Final Dimensions—Described in Chapter 3
45
8x
4.750
8x
.25 THRU
.250 THRU
2x
SCALE 0.500
.500 THRU
22.500°
WINDOW CENTER = 0
1.875
4.500
APPROVED
CHECKED
DRAWN
R.S.
4/27/2017
DWG NO.
0.5
A3
SIZE
SCALE
TITLE
ROTOR SHELL TOP
FINAL DIMENTIONS
3
5 OF 5
UNIVERSITY OF ILLINOIS
TOLERANCE
0.00 = 0.01
0.001 = 0.003
REV
Figure A.9: Rotor Shell Front Air Hole Final Dimensions—described in
Chapter 3
46
.400
R.200
.544
.272
1.168 REF
.980
.620
.375
.245 REF
1.000
-B-
APPROVED
CHECKED
DRAWN
R.S.
5/9/2017
DWG NO.
0.143
A3
SIZE
SCALE
TITLE
ROTOR SHELL FRONT
FAN HOLES
3 OF 5
UNIVERSITY OF ILLINOIS
SECOND THESE HOLES ARE DRILLED
ONE OVER, WHICH IS THE UPPER
DIMENTIONS
THESE HOLES ARE DRILLED TWICE
FIRST THESE HOLES ARE DRILLED
.544X1.000
REV
Figure A.10: Fan Front View—Creates Air Flow
47
.02
14.00°
65.00°
28 FINS MACHINED FROM STEP FILE
TOLERANCE
.00 = 0.01
.000 = 0.003
R3/32
RADIUS ON FINS
B
B
Ø11.730
APPROVED
CHECKED
DRAWN
R.S.
5/1/2017
DWG NO.
0.5
1 OF 2
UNIVERSITY OF ILLINOIS
REV
Ø11.730
Ø7.700
SECTION B-B
SIZE
A3
.300
R.25
Ø6.500
SCALE
TITLE
FAN FRONT
.700
R1.00
.400
.900
.700
Figure A.11: Fan Back View—Creates Air Flow
48
8x
.250
4.750
22.500°
SCALE 0.500
2x
12.500°
.500
4.500
APPROVED
CHECKED
DRAWN
R.S.
5/1/2017
.40
DWG NO.
0.5
A3
SIZE
2 OF 2
UNIVERSITY OF ILLINOIS
SCALE
TITLE
FAN BACK
REV
Figure A.12: Fan Lid—Eliminates Air Leakage between Fan Channels
49
11.210
11.730
SLIP FIT TO EDGE OF SHELL
R5.741
28x #36 HOLES
.200
TOLERANCE
APPROVED
CHECKED .00 .01
.000 .003
DRAWN
R.S.
5/5/2017
SCALE
A3
SIZE
TITLE
FAN LID
0.250
DWG NO.
1 OF
UNIVERSITY OF ILLINOIS
REV
Figure A.13: Carbon Fiber Ring—Restrains All Parts, Reduces Stress in
All Parts
50
13.550
12.530
TOLERANCE
APPROVED
CHECKED .00 .01
.000 .003
DRAWN
Reed
10.100
SCALE
A3
SIZE
0.167
DWG NO.
1 OF
UNIVERSITY OF ILLINOIS
TITLE
CARBON FIBER RING FINAL
REV
Figure A.14: Lock Nut—Sets Preload on Bearings, Keeps All as One Unit
51
A
A
6x
.250
.350
2.000
.500
4.02
R2.335
.050
1.688
THREAD SIZE
3.375" MAJOR
3.275" MINOR
16 THREADS/in
LEFT HAND THREAD
TOLERANCE
APPROVED
CHECKED .00 .01
.000 .003
DRAWN
R.S.
5/3/2017
DWG NO.
1.000
A3
SIZE
SCALE
1 OF 1
UNIVERSITY OF ILLINOIS
TITLE
LOCK NUT FINAL
SECTION A-A
4.671
.350 .400
R.100
REV
Figure A.15: Balancing Ring —Restrains Magnets, is Another Surface to
Use for Balancing for Rotordynamics
52
MAKE FROM Ti-6Al-4V
.490
.250
1/8
.010
DETAIL A
SCALE 4.000
.30
.20
0.25 FOR BALANCING, MAXIMUM
FOR REFERENCE
.032
.065
.194
.227
12.160
11.160
APPROVED
CHECKED
DRAWN
R.S.
5/2/2017
.00 .01
.000 .003
TOLERANCE
DWG NO.
.5
A3
SIZE
SCALE
1 OF 1
UNIVERSITY OF ILLINOIS
12.160
TITLE
BALANCING RING
SEE DETAIL A
12.140
REV