Professional Documents
Culture Documents
Hallin Thesis Final File
Hallin Thesis Final File
Princeton, NJ
Statement on Copyright Restrictions
This senior thesis can only be used for education and research (among other purposes consistent
with Fair use) as per U.S. Copyright law (text below). By accessing this file, all users agree that
their use falls within fair use as defined by the copyright law. They further agree to request
permission of the Princeton University Library (and pay any fees, if applicable) if they plan to
publish, broadcast, or otherwise disseminate this material. This includes all forms of electronic
distribution.
U.S. Copyright Law (Title 17, United States Code)
The copyright law of the United States governs the making of photocopies or other
reproductions of copyrighted material. Under certain conditions specified in the law,
libraries and archives are authorized to furnish a photocopy or other reproduction. One of
these specified conditions is that the photocopy or other reproduction is not to be used for
any purpose other than private study, scholarship or research. If a user makes a request
for, or later uses, a photocopy or other reproduction for purposes in excess of fair use, that
user may be liable for copyright infringement.
Submitted to the
Department of Mechanical and Aerospace Engineering
Princeton University
in partial fulfillment of the requirements of
Undergraduate Independent Work.
Final Report
May 1, 2014
Professor Nosenchuck
Professor Hultmark
MAE 442
160 pages
File Copy
Abstract
Elon Musk captured the worlds attention in August 2013 when he released a highlevel alpha design for a fifth mode of transportation called the Hyperloopa reducedpressure tube that contains pressurized capsules driven within the tube by a number
of linear electric motors. His proposal was motivated by the proposed high speed
rail system between Los Angeles and San Francisco that would cost $68.4 billion and
have a total trip time of 2 hours and 38 minutes, providing limited transportation
improvements. The alpha design outlines a passenger system between the same two
cities that could be built for only $6 billion and have a faster travel time of only 35
minutes, offering significant advancements.
While this alpha document provided a complete overview with clear goals for the
system, more thorough engineering analysis was required to assess its technical viability as well as its economical reality. Each subsystem was broken down into its
fundamental parameters and governing physical principles. These were then analyzed to evaluate their effect on both their corresponding subsystem and the overall
Hyperloop performance. Several modifications were suggested to make the system
feasible. Issues downplayed by the alpha document were brought to attention and
their impacts were discussed. Economics, politics, and other human factors were also
considered to complement the engineering analysis to gain a top-level perspective on
this new mode of transportation. From the analysis it was determined that the Hyperloop could make the journey in 36.35 minutes with a revised capital expenditure
of $16.84 billion. The Hyperloop would be recommended as an alternative to the high
speed rail and has the potential to revolutionize transportation across the world.
iii
Acknowledgements
I would like to thank my adviser, Professor Nosenchuck, for all his support, wisdom,
time, and energy throughout the senior thesis. It was rewarding working with someone
with so much experience and knowledge as I was encouraged to explore areas of
engineering I would have never before considered.
In addition, without the guidance and teaching from the entire Mechanical and
Aerospace Department faculty and staff I would have never developed the tool set to
take on such a complex project that reaches across all disciplines of the department.
iv
Contents
Abstract . . . . . .
Acknowledgements
List of Tables . . .
List of Figures . . .
List of Symbols . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
iii
iv
x
xi
xv
.
.
.
.
.
1
1
1
2
3
5
.
.
.
.
.
.
.
.
.
.
.
6
6
6
7
8
9
9
10
10
11
11
11
14
14
14
1 Introduction
1.1 Background and Significance . . . . . . . . . . . .
1.1.1 Motivation Behind Elon Musks Proposal .
1.1.2 Seeking an Alternative to High Speed Rail
1.1.3 Similar Proposals to the Hyperloop . . . .
1.2 Objectives . . . . . . . . . . . . . . . . . . . . . .
2 System Overview
2.1 Technical Components and Principles . . . .
2.1.1 Aerodynamic Drag . . . . . . . . . .
2.1.2 Capsule Propulsion System . . . . .
2.1.3 Capsule Suspension System . . . . .
2.1.4 Vacuum Pumps . . . . . . . . . . . .
2.1.5 Compressor . . . . . . . . . . . . . .
2.1.6 Hyperloop Route . . . . . . . . . . .
2.1.7 Infrastructure . . . . . . . . . . . . .
2.2 Total Cost Overview . . . . . . . . . . . . .
2.2.1 Estimated Capsule Cost and Weight
2.2.2 Total System Capital Costs . . . . .
vi
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3.3
3.4
3.5
3.6
3.7
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
15
16
17
18
19
20
23
26
.
.
.
.
.
.
.
.
.
28
28
28
28
30
31
32
33
35
37
.
.
.
.
.
.
.
.
.
.
40
40
40
42
42
42
43
45
49
50
54
58
58
59
61
6.4
6.5
6.6
6.7
6.8
Preliminary Analysis . . . .
H Control Synthesis . . . .
Robust Synthesis . . . . .
Other Papers/Other Control
Summary . . . . . . . . . .
. . . . . .
. . . . . .
. . . . . .
Methods .
. . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
62
64
69
72
72
73
73
73
73
74
74
75
8 Hyperloop Compressor
8.1 Introduction . . . . . . . . . . . . . . .
8.2 Alpha Compressor Design . . . . . . .
8.3 Diffuser Compressor Design . . . . . .
8.4 Intercoolers: Water Coolant Mass Flow
8.5 Summary . . . . . . . . . . . . . . . .
86
86
86
87
89
91
. . .
. . .
. . .
Rate
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
75
79
80
80
85
9.4
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
10 Hyperloop Infrastructure
10.1 Tube Route Layout Schemes . . . . . . . . . . . . . . . . . . . . . .
10.2 Tube Material and Structure . . . . . . . . . . . . . . . . . . . . . .
10.3 Tube Thickness Numerical Analysis . . . . . . . . . . . . . . . . . .
10.4 Pylons, Expansion Joints, and Dampers . . . . . . . . . . . . . . .
10.5 Large-Scale Engineering and Infrastructure Projects . . . . . . . . .
10.5.1 Princeton Plasma Physics Laboratory and Engineering Costs
10.5.2 Other Large-Scale Infrastructure Project Costs . . . . . . . .
10.6 Hyperloop Infrastructure Costs . . . . . . . . . . . . . . . . . . . .
10.6.1 Tube Costs . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.6.2 Pylon Construction Costs . . . . . . . . . . . . . . . . . . .
10.6.3 Tunneling Costs . . . . . . . . . . . . . . . . . . . . . . . . .
10.7 Total Civil Infrastructure Costs Summary . . . . . . . . . . . . . .
11 Effect of Changing System Parameters on Overall
System Costs
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . .
11.2 Effect of Changing Tube Diameter . . . . . . . . . . .
11.3 Effect of Changing Capsule Length . . . . . . . . . .
11.4 Effect of Changing Tube Pressure . . . . . . . . . . .
11.5 Effect of Changing Suspension Gap Height . . . . . .
11.6 Other effects . . . . . . . . . . . . . . . . . . . . . . .
11.7 Total Costs . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
109
109
110
110
112
113
113
115
116
116
116
118
119
12 Conclusion
133
143
ix
List of Tables
2.1
2.2
2.3
4.1
4.2
4.3
4.4
11
12
13
33
35
37
38
5.1
5.2
5.3
41
45
57
6.1
6.2
61
66
7.1
. . . . . . . . . . .
75
9.1
9.2
9.3
96
96
99
List of Figures
1.1
2.1
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
16
20
21
22
22
23
25
26
4.1
29
31
36
37
4.2
4.3
4.4
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
5.9
xi
41
44
44
46
48
48
49
51
52
53
54
56
59
62
63
64
65
7.1
7.2
7.3
7.4
. . .
. . .
. . .
0 to
. . .
. . .
. . .
. . .
. . .
. . .
75
76
77
7.5
7.6
7.7
7.8
7.9
8.1
87
9.1
9.2
9.3
93
94
95
xii
. . . .
. . . .
. . . .
Range
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
67
68
69
70
71
78
79
81
82
83
84
9.4
9.5
9.6
9.7
9.8
9.9
9.10
9.11
9.12
9.13
9.14
95
98
98
100
102
102
104
105
105
106
107
xiii
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
121
122
123
124
125
126
127
127
128
129
131
List of Symbols
vc
Speed of Sound . . . . . . . . .
Dynamic Viscosity . . . . . . .
Dtube Diameter of Hyperloop Tube .
hgap
Suspension Gap Height . . . .
Wcapsule Width of Capsule . . . . . . .
Hcapsule Height of Capsule . . . . . . .
Lcapsule Length of Capsule . . . . . . .
br
Blockage Ratio . . . . . . . . .
SA
Capsule Frontal Surface Area .
SA0
Tube Cross-Sectional Area . .
F
Force . . . . . . . . . . . . . .
Shear Stress . . . . . . . . . .
p
Pressure . . . . . . . . . . . .
gf orce g-force . . . . . . . . . . . . .
t
Time . . . . . . . . . . . . . .
mc
Mass of Capsule . . . . . . . .
W
Work . . . . . . . . . . . . . .
P
Power . . . . . . . . . . . . . .
Efficiency . . . . . . . . . . . .
Wski
Air Suspension Ski Width . . .
Lski
Air Suspension Ski Height . . .
Nski
Number of Air Suspension Skis
xiv
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
15
15
15
15
15
15
15
15
16
16
16
16
16
16
16
16
17
19
19
32
32
33
33
33
34
41
41
41
V
Kn
C
M
Se
S0
r
rp
m
cp
Q
rmin
xv
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 41
. 42
. 43
. 74
. 74
. 74
. 74
. 88
. 88
. 88
. 88
. 89
. 94
. 111
Chapter 1
Introduction
1.1
1.1.1
Elon Musk captured the worlds attention in August 2013 when he released a highlevel alpha design for a fifth mode of transportation called the Hyperloopa reduced
pressure tube that contains pressurized capsules driven within the tube by a number
of linear electric motors. His proposal was motivated by Californias proposed high
speed rail system between Los Angeles and San Francisco that would cost $68.4
billion, have a travel time of 2 hours 38 minutes, a $105 one-way ticket price, and
high energy costs [34]. Disappointed, Musk established the following objectives for
the Hyperloop:
Lower cost
Faster
Safer
More convenient
Immune to weather
Sustainably self-powering/Environmentally friendly
Resistant to earthquakes
Not disruptive to those along the route
1
The alpha design outlines a passenger system between the same two cities that could
be built for only $6 billion, achieve a faster travel time of only 35 min, have cheaper
tickets of $20, and remove human control error and unpredictable weather from the
system, while managing to satisfy every other objective on this list. The Hyperloop
seeks to revolutionize transportation, but major questions remain about the possibility of actually building this system and meeting these objectives.
1.1.2
Any transportation system connecting two major cities such as L.A. and San Francisco
will require a massive investment that will need to be met with returns exceeding
expectations. Californias high speed rail system does not seem to warrant this
investment. The proposed rail would be one of the most expensive per kilometer
and slowest in the world. It is not only more expensive to operate than planes if
unsubsidized, but also slower and less safe than flying by two orders of magnitude
(based on estimates from the Bureau of Transportation Statistics [59]). The returns
of the proposed high speed rail are minimal when it does not reduce current trip times
or reduce the cost relative to existing modes of transportation. The Hyperloop offers
a promising alternative.
There are several environmental advantages of the Hyperloop. A preliminary
energy cost analysis presented in the alpha document (Figure 1.1) further reveals
substantially higher costs for the high speed rail compared to that of proposed Hyperloop. In addition, trains require wide sloths of land and are also loud, require
fences, can be safety hazards to others, and have a higher risk of derailment. The
noise level of the Hyperloop will be virtually silent compared to the damaging noise
levels of trains and the Hyperloop will produce negligible ground vibration. Furthermore, there is no legislation to prevent the Hyperloops implementation and the
California environmental laws that will ramp up in 2020 will greatly favor this new
mode of transportation. In January 2014, Nick Garzilli submitted a request to place
the Transportation Innovation Act on the California ballot. This act would suspend
further issuance of bonds and construction of the high speed rail system in order to
allow for the construction and operation of innovative technologies that are faster and
more reliable, energy efficient, flexible, and less expensive [20].
Figure 1.1: Energy cost per passenger for a journey between Los Angeles and San
Francisco for various modes of transport [34].
The Hyperloop may not be the right solution for all travel distances. Musk discusses an inflection point where for high traffic city pairs less than 1500 km apart the
Hyperloop is the right solution, but for greater distances hypersonic air travel makes
more sense.
1.1.3
Prior to Musks Hyperloop proposal, Robert Goddard, Rand Corporation, and ET3
have presented similar ideas. Many use the term evacuated tube transportation
(ETT) to describe this transportation technology.
Robert Goddard, the revolutionary rocket scientist, first conceptualized vac-trains
in the 1910s. These trains would levitate via magnets and move along a partial
vacuum tube at very high speeds due to lack of air resistance [19].
In 1972, Rand Corporation published a paper proposing the Very High Speed
Transit System (VHSTS), a maglev-based underground vacuum tunnel system [48].
3
A main line would run diagonally across the country and could make the trip between
New York and L.A. in only 21 minutes while achieving unrealistic speeds of 14,000
mph.
Daryl Oster formed the company Evacuated Tube Transport Technologies (ET3)
to pursue his idea for an ETT system with the claim of transporting passengers
between New York and L.A. in only 45 minutes, and ultimately sending passengers
and cargo from New York to Beijing in 2 hours [38]. Like Musk, Oster was dissatisfied
with current transportation dependencies and has called for a new paradigm shift in
transportation. Transportation has the potential to be the biggest growth market in
the world and his global vision would connect the world with a network of evacuated
tubes within the next 20 years. He first conceived the idea in the 1980s and later
received his first patent in 1999. The initial designs would transport 6 passengers or
367 kg of cargo in a system of vacuum tubes at speeds of 600 km/hr, but eventually be
propelled via maglev up to 6,500 km/hr (Mach 5) in straight unpopulated areas. This
really embodies ET3 trademarked phrase Space Travel on Earth as this hypothetical
top speed nearly doubles the fastest recorded airspeed record. While Osters proposed
$10 $20 trillion investment to displace 90% of the worlds current transportation
scheme is extraordinarily ambitious and unrealistic, the company is currently focusing
on finding 5 km stretches of land to test the elements of their concept needed to
achieve 600 km/hr. ET3 has estimated this test track will cost $20 million, but upon
large-scale implementation the costs will drop to $1.86 million per km. Oster has
also claimed to be able to build his system between L.A. and San Francisco for 1/10
the cost of the proposed high speed rail as well as 1/50 the energy. The ET3 website
has a table comparing ET3 costs with high speed rail costs [37]. However, neither
Oster nor the other 244 contributing ET3 licensees have released any design plans or
detailed cost estimates to back these claims.
The Rand and ET3 designs are centered around drawing hard or near hard vacuum
in the tube and using electromagnetic suspension. This is problematic as it is very
difficult to maintain the vacuum they seek. An alternative is to have a low pressure
system set to a level where standard commercial pumps can easily overcome an air
leak and the transport capsules can handle variable air density. This system would
be inherently robust. To avoid the high costs of maglev suspension technology, an
air suspension system could be implemented. Modifications to these earlier ideas,
provide the basis for the Hyperloop concept as a new method of public transit.
1.2
Objectives
This thesis aims to provide a detailed analysis of the Hyperloop defined by Elon
Musks alpha document. The existing literature was surveyed to gain insight into the
individual subsystems of the Hyperloop. While the literature is replete with interest,
many engage in hand waving without justification. More detailed engineering analysis
was necessary to determine the technical viability of this system. The objective
was to then individually break down each subsystem and to develop performance
and cost metrics to bring all the elements together. These metrics could then be
used to compare to other transportation systems. Complexities of the design were
simplified when necessary and modifications were suggested to make the concept
realizable. Economics, politics, and other human factors were also considered to
complement the engineering analysis to gain a top-level perspective on this new mode
of transportation.
Chapter 2
System Overview
The Hyperloop is a single system that incorporates the tube, vehicle, propulsion
system, suspension system, energy management, timing, and route. There is the possibility to build a passenger plus vehicle Hyperloop, but the main focus here will be
on the passenger-only concept. The following is an overview of the Hyperloop components as defined by the alpha document. The alpha outline provided the basis for
a top-down systems engineering analysis. Each subsystem was broken down into its
fundamental parameters and governing physical principles. These were then analyzed
to evaluate their effect on both the corresponding subsystem and the overall Hyperloop performance. Appendix A should be consulted for the important parameters of
the Hyperloop system defined in the alpha document. These parameters were used
to develop a detailed discussion of the Hyperloop in the following chapters.
2.1
2.1.1
Tube Pressure
The Hyperloop capsule will be transported in a partially evacuated cylindrical tube
supported by pillars in a closed system. For a capsule moving in a tube with air, the
greatest power requirement is to overcome air resistance. Since aerodynamic drag
increases with the square of the speed, the power requirement increases with the
cube of the speed. To reduce the drag force and manage shock waves as the capsule
approaches the speed of sound, the operating tube pressure is set to 100 Pa (1000
6
times less than sea level conditions). Further reducing the tube pressure would be
offset by increased pumping complexity (see Chapter 7).
Tube and Capsule Geometry
The aerodynamics of the system require the capsule to tube area ratio (blockage ratio)
to be as small as possible in order to best reduce the risk of choking the flow and to
minimize the amount of power required to reach and maintain capsule speed. The
capsule must displace its own volume of air as it moves through the tube and any flow
that is not displaced may accelerate to supersonic velocities through the constricted
gaps between the capsule and the tube, forming shock waves. Choking the flow and
the formation of shock waves should be avoided as the required power will increase
significantly to overcome the increased drag and additional mass of air in front of
the vehicle. A compressor mounted on the capsules leading face will ingest air that
is not displaced, mitigating these adverse effects. In addition, the capsules speed is
limited to 760 mph in order to keep the flow subsonic for the given tube conditions.
The geometry of the capsule is streamlined to reduce drag (Figure 2.1). The frontal
area is optimized to maximize speed and performance while maintaining passenger
comfort. This corresponds to maximum width of 1.35 m and maximum height of
1.10 m. Not including any propulsion or suspension components this is equivalent to
a frontal area of 1.40 m2 . The alpha documents optimized inner tube diameter is
2.23 m with a corresponding tube cross-sectional area of 3.91 m2 and a blockage ratio
of 36%.
2.1.2
An advanced magnetic linear motor system will be developed to propel the vehicle
to travel speeds. The system will be required to maintain passenger comfort while
accelerating the capsule to 760 mph (339.75 m/s). To achieve this, the max acceleration is around 1g. For roughly 90% of the journey the capsule coasts and will not
require propulsion. There will be several major propulsion stations along the journey
to get the capsule up to speed as well as various reboost stations to maintain capsule
speed. The tube will have a stationary element (stator) which powers the vehicle.
The capsule will contain the moving rotor element (rotor). The rotors located on the
capsule transfer momentum to the capsules via linear accelerators. The Hyperloop
will be self-powering, using regenerative braking and solar power to reduce the energy requirements. A solar array will cover the entire tube with an expected energy
7
2.1.3
2.1.4
Vacuum Pumps
Vacuum pumps are necessary to first evacuate air from the Hyperloop tube before
operation and to then maintain the desired pressure in the presence of any possible
leaks throughout the tube. These pumps will run continuously along the length of
the tube. Vacuum pumps will be selected based on pump-down time requirements.
Any stations, emergency exits, or branches along the Hyperloop will be isolated from
the main tube to minimize air leaks.
2.1.5
Compressor
A factor limiting the high speed movement of the Hyperloop capsule through a tube
containing air is the Kantrowitz limit [61]. For a given tube to pod ratio there is
a maximum speed that will choke the flow. If the capsule is too close to the tubes
walls then the capsule will eventually be forced to push the entire column of air in the
system. The Kantrowitz limit constrains the system to either go slowly or have a very
large tube diameter, neither of which are ideal. The proposed solution is to mount an
electric compressor fan on the nose of the pod that will actively transfer high pressure
air from the front to the rear of the capsule. Throughout the journey the weight of
the capsule will be supported by an air bearing system and this compressor will also
serve to supply the necessary pressurized air.
2.1.6
Hyperloop Route
The proposed alpha route from Los Angeles to San Francisco will follow I-5 and I580 for the majority of the 563 km journey. The minimum bend radius to maintain
passenger comfort will dictate the speed of the capsule in each section of the route.
The three cruising speeds of the capsule are 300 mph (134.11 m/s), 550 mph
(248.11 m/s), and 760 mph (339.75 m/s). Propulsion stations are used to accelerate
the capsule to each of these speeds and regenerative braking stations are then used
to decelerate the vehicle. The journey will take 35 minutes one way. Capsules will
be separated by 37 kilometers or depart roughly every 2 minutes. The passenger
Hyperloop can carry up to 28 people and will transport 7.4 million people per year.
2.1.7
Infrastructure
Tube Construction
The inner diameter of the tube will be constructed from a uniform thickness steel
tube reinforced with stringers. To provide sufficient strength for such load cases as
pressure differential, bending and buckling between pillars, loading due to the capsule
weight and acceleration, and seismic conditions the tube thickness should be between
20 mm and 23 mm. The inside of the tube will be finished to a smooth gliding surface
using a specifically designed cleaning and boring machine.
Pylon Construction
The tube will be built on reinforced concrete pylons 6 m tall placed every 30 m to
keep material cost and deflection to a minimum. They will not be rigidly fixed as
there are two adjustable lateral (XY) dampers and one vertical (Z) damper. They
are constructed to withstand earthquakes and have expansion joints. Slip joints at
stations will be able to handle the tube length variance due to thermal expansion.
The tube is built above ground on pylons which allows prefabricated sections to
be dropped in place. The tube can follow I-5 to avoid buying large amounts of land.
10
2.2
2.2.1
The alpha documents estimations for an individual capsules cost and mass can be
found in Table 2.1.
2.2.2
Two one-way tubes plus 40 capsules will cost $6 billion incorporating a $0.5 billion
cost margin (Table 2.2). The capsules are only 1% of the total budget. The tubes
represent 70% of the budget, or $4.06 billion. The estimated total cost is only 9% of
the California high speed rail. The service life will be 100 years.
A passenger plus transport version (Table 2.3) will only cost 25% more (still only
11% of the cost of the high speed rail).
11
12
Table 2.3: Estimated total cost of Hyperloop passenger plus vehicle transportation
system.
Component
Cost (million USD)
Cargo Capsule
30.5 (20 capsules)
Capsule Structure & Doors
5.5
Interior & Seats
3.7
Compressor & Plumbing
6.0
Batteries, Motor & Electronics
4.0
Propulsion
3.0
Suspension & Air Bearings
5.3
Components Assembly
3.0
Passenger Only Capsule
40.5 (30 capsules)
Capsule Structure & Doors
7.4
Interior & Seats
7.6
Compressor & Plumbing
8.2
Batteries, Motor & Electronics
4.5
Propulsion
3.8
Suspension & Air Bearings
6.0
Components Assembly
3.0
Tube
7,000
Tube Construction
1,200
Pylon Construction
3,150
Tunnel Construction
700
Propulsion
200
Solar Panels & Batteries
490
Station & Vacuum Pumps
260
Permits & Land
1,000
Cost Margin
429
Total
7,500
13
Chapter 3
Aerodynamic Analysis of the
Hyperloop
3.1
Overview
The aerodynamic drag acting on the Hyperloop capsule as it travels through the
tube depends on many factors. Numerical simulations using MATLAB1 were used to
investigate the effect of tube air pressure, capsule speed, capsule shape, and blockage
ratio. Appendix A should be consulted for the important parameters of the Hyperloop
system defined in the alpha document.
3.2
Introduction
Traditional trains are constrained to low speeds because a large percentage of the
system energy is lost to the dynamic friction between the wheels of the train and
the track rail. Maglev trains were introduced and designed to avoid this friction by
floating above the track using magnetic levitation technology. Despite improvements,
the velocity of the trains only achieved maximum commercial speeds of 400 - 500
km/hr. At these speeds aerodynamic drag is very high, about 80-90% of the total
drag [29]. If drag could be reduced, then the speed of the vehicle could be increased.
Evacuated tube transport (ETT) provides a solution to this problem by substantially
reducing the pressure in the tube. In theory, an absolute vacuum would have zero
1
MATLAB was used for the analyses, results, and figures seen throughout all chapters
14
aerodynamic drag. While this is not feasible, low pressures 1000 times lower than
atmospheric pressure can be achieved using current vacuum pump technology (see
Chapter 7) and the shape of the vehicle can be further optimized to achieve very low
drag forces. Simulations using the Navier-Stokes equation of compressible viscous
flow and turbulence models such as k- can be implemented. To study the effect of
aerodynamic drag in the evacuated tube of the Hyperloop, the basic mass conservation
and momentum conservation equations were used to develop expressions for the total
drag on the Hyperloop capsule as a function of desired parameters.
3.3
The flow environment of the tube is very important and the following parameters can
help evaluate the flow.
The speed of sound for a calorically perfect gas is given by 2
vc =
RT
(3.3.1)
where is the ratio of specific heats, R is the specific gas constant, and T is the fluid
temperature.
The Mach number helps to characterize the flow regime. For a fluid with velocity
ue , the Mach number is defined as
M=
ue
vc
(3.3.2)
The Reynolds number is a measure of the ratio of inertia forces to viscous forces.
This powerful non-dimensional parameter is helpful for predicting laminar or turbulent flow, and it is defined as
ue h
(3.3.3)
Re =
where is the fluids density, is the fluids dynamic viscosity, and h is the characteristic length.
For a fluid flow of ue, max = 339.75 m/s and a tube temperature of T = 293.15
K, the speed of sound is vc = 343.20 m/s and the Mach number is M = 0.99. Thus,
at the alpha max speed the flow is just subsonic. Special attention needs to be paid
to speed of sound locally, the effect of shock loss, and pressure recovery.
2
vc is used instead of the conventional symbol a to avoid confusion with other variables
15
3.4
The geometries of the tube and capsule were simplified in order to observe overall
trends of characteristic parameters. In Figure 3.1 the width and height of the tube
are defined by a0 , b0 , respectively; the width, height, and length of the capsule are a,
b, c, respectively.3 To make the model approximation consistent with the dimensions
of the tube a0 = b0 = 0.785 Dtube (this makes the cross sectional area of the square
tube model equal to that of the circular pipe with diameter Dtube ). The capsule is
suspended above the bottom the tube by h (suspension gap height).
The blockage ratio, br , is defined by
SA
ab
=
a0 b 0
SA0
(3.4.1)
where SA and SA0 are the frontal surface area of the capsule and the yz-cross sectional
surface area of the tube, respectively.
16
Examining the mass and momentum conservation equations in the x-direction for an
infinitesimal fluid element
(x, t)
+ (vx ) = 0
t
(3.4.2)
[(x, t) vx ]
p
+ [(x, t) vx2 ] = (vx )
+ fx
t
t
(3.4.3)
where is the airs density in the tube, vx is the capsules velocity in the x-direction4 ,
is the dynamic viscosity of the air in the tube, and fx is the external force per
unit volume acting on the fluid element in the x-direction. A streamlined nose of the
capsule is not considered in this derivation in order to just highlight the effects of
pressure in the tube, capsule speed, and blockage ratio.
There are three components of the capsules drag that are considered.5
1. F1 : The force on the front of the capsule due to the collision between air and
the vehicle (only for velocities greater than the speed of sound)
2. F2 : The air friction on the four faces of the capsule
3. F3 : The force caused by the pressure differential between the front and rear of
the capsule
3.4.1
F1 Derivation
After air collides with the front of the capsule it will displace a small layer a short
distance dx away in a small time dt. Thus, the velocity of this thin layer of air will
change dt after the collision. The velocity of the infinitesimal air before the collision
is equal to zero. Its velocity after the collision is equal to the capsules. The kinetic
energy of the air can be equated to the force differential times the distance traveled
following the collision.
1
(3.4.4)
dxdydz vx2 = dF1x dx
2
Due to Brownian motion F1 is equal to zero when the velocity of the capsule is less
than the speed of sound (vc ). When the velocity is greater than vc the air column with
length vc dt is affected such that its velocity is approximately equal to the capsules
velocity. Combining these formulations with the momentum equation yields
4
5
17
0
RR
F1 =
vc
yz
(
p
2p0 0
vx dydz =
0
br SA0
2p0 0
vc vx
vx < vc
vx vc
(3.4.5)
3.4.2
F2 Derivation
Friction force due to air is generated on the four faces of the capsule. The fluid flow
between all four faces and the tube wall will be approximated as Couette flow. This
flow is defined as laminar flow of a viscous fluid in the space between two parallel
plates, one of which is moving relative to the other. The tube wall is treated as
the fixed surface and the capsules face can be represented by a moving plate at the
freestream velocity, ue = vx . Due to the no-slip condition, there can be no relative
motion between the plate and the fluid. The moving plate exerts a shear stress on
the fluid causing the fluid to move. Couette flow is a constant pressure flow.
Considering an infinitesimal volume on the capsules upper surface, the velocity
profile is a function of z (vx0 = f (z)) and the area of contact is ds = dx dy. From the
x-momentum equation
vx0
(
)=0
(3.4.6)
z z
And for constant viscosity, ,
2 vx0
=0
2z
(3.4.7)
z
b b0 h
(3.4.8)
(3.4.9)
The shear stress acting between the capsule and the tube wall is given by the rela-
18
vx0
z
(3.4.10)
The shear force can then be integrated over the area of contact,
F2U = tube dxdy
(3.4.11)
The friction forces acting on the other three surfaces can be determined in a similar
manner and thus,
F2 = F2U + F2B + F2L + F2R
(3.4.12)
The bottom gap is very small and a temperature rise will be observed, resulting in a
different viscosity (gap ) from the surrounding tube (tube ). Recognizing this viscosity
difference and evaluating the integrals over the defined surface of each face results in
F2 = cvx tube
3.4.3
a
2b
+ 1
b0 b h
(a a0 )
2
+ gap
a
h
(3.4.13)
F3 Derivation
There will be a pressure differential between the front and rear of the Hyperloop
capsule as it moves through the tube (Figure 3.2). The pressure inside the tube is
p. Using the capsule as reference, the velocity of the infinitesimal fluid element is vx .
From Bernoullis formula, the pressure at point A is defined as
1
p31 = p + vx2
2
(3.4.14)
and the pressure at point B, neglecting any additional pressure due to the thrust
gained from the exit of the onboard compressor, is
p32 = p
(3.4.15)
19
1 2
v dydz
2 x
(3.4.16)
p 0 2
v
2p0 x
(3.4.17)
The total drag force acting on the capsule in the x-direction is then
Fdrag, x = F1 + F2 + F3
(3.4.18)
3.5
Numerical Analysis
Using MATLAB, plots were produced to demonstrate the effect of changing pressure,
velocity, and blockage ratio on the total aerodynamic drag.
Figure 3.3 and 3.4 demonstrate the effect of increasing pressure and capsule velocity. In Figure 3.3, the velocity ranges from 0 to 400 m/s and the pressure ranges from
0 to 10,000 Pa. From the bottom leftmost plot it is evident that F1 only has an effect
once the capsule speed exceeds the speed of sound (vc = 343.2 m/s). In addition, this
force increases with both velocity and pressure. Despite not accurately modeling the
impact of shocks once the capsules travel becomes supersonic, the negative effects
of reaching these speeds can still be observed in the large step increase in the total
aerodynamic drag (top plot) due to F1 . Only subsonic travel is considered for the
Hyperloop and thus, F1 will not be considered beyond this point. The plot of the air
friction contribution to drag reveals a linear increase with the speed and no pressure
dependence. Force due to the pressure differential is impacted more drastically by
20
the increase in pressure and speed. While the friction forces only reach levels of 102
Newtons, the pressure differential force rises to 105 Newtons. Clearly, to reduce drag
the contribution due to pressure differential needs to be decreased by evacuating the
tube to a low pressure.
Figure 3.3: Aerodynamic Drag as Function of Tube Pressure and Capsule Velocity.
The top plot is the total aerodynamic drag acting on the Hyperloop capsule. The three plots beneath
show the contours for the three force components. The leftmost plot shows F1, the center plot shows
F2, and the right plot shows F3.
Figure 3.4 examines the aerodynamic drag relationship with pressure and velocity
only in pressure ranges that would be implemented in the Hyperloop. From the
surface plot it is evident the drag force due to pressure is substantially less. It is
only 800 N at maximum speed and 600 Pa (6 times the tube pressure proposed by
the Hyperloop). The total drag force is substantially reduced. Figure 3.5 shows the
aerodynamic drag as a function of pressure for the alpha document conditions. For
the Hyperloops 100 Pa environment and at max velocity of 339.75 m/s, the total
drag is only 362 N, which is at a magnitude that will allow the capsule to coast for
most of its journey. The aerodynamic drag forces effect on the capsules speed and
position as a function of trip time can be examined in Chapter 9. Reducing the
pressure to 0 Pa will give the lowest obtainable drag, but this is not feasible. For all
three alpha speeds, further reducing the pressure below 40 Pa does not reduce drag
by a substantial amount to warrant using more complex vacuum pump stations to
achieve a lower pressure.
21
Figure 3.4: Aerodynamic Drag as a Function of Tube Pressure and Capsule Velocity.
The drag is substantially lower when the pressure range is narrowed down to 0 to 600
Pa.
Total Aerodynamic Drag Force vs. Tube Pressure:
Alpha Conditions
500
450
400
350
300
250
ue,max
ue,mid
200
ue,low
150
100
50
0
50
100
Pressure (Pa)
150
200
Figure 3.5: Aerodynamic drag as a function of tube pressure for alpha document
conditions. Suspension gap = 1.3mm and ue, max = 339.75 m/s, ue, mid = 248.11 m/s,
and ue, low = 134.11 m/s
22
The effect of blockage ratio was then examined. The following were held constant
and equal to alpha document specifications: all tube conditions, all bottom gap conditions (gap height = 1.3mm), the ratio of the capsules width to height, the capsules
length, and the capsules velocity (taken to be ue,max ). Figure 3.6 demonstrates by
increasing the blockage ratio the drag force increases. The drag force appears to start
leveling off as br approaches 0.7. However, there is a point where the surface area of
the capsule will be large enough to choke the flow at the Hyperloops high speeds.
Once the flow has become choked, air will build up in front of the capsule and the
aerodynamic drag will dramatically increase. This is not demonstrated as part of this
particular analysis.
Aerodynamic Drag Force vs. Blockage Ratio
600
500
400
300
200
100
0
0
0.1
0.2
0.3
0.4
Blockage Ratio
0.5
0.6
0.7
3.6
External Studies
Several external studies have looked into the aerodynamic effects of a vehicle in an
evacuated tube. Zhang (2011) considered a 40 meter subsonic Maglev train in an
evacuated tube in feasible vacuum pressures of 1 to 1000 Pa. The research suggested
the optimal blockage ratio for this form of ETT should be in the range of 0.25 to 0.70
and the tube diameter to be 2-4 meters [66].
23
Figure 3.7: Ansys analysis of the Hyperloop: contours of wall shear stress (Pa).
Ansys has further suggested that adding air bearings to the top of the capsule would
do a better job of spreading air across the body and help the vehicle stay balanced
during slight changes in air pressure.
The simulation also revealed an issue with the amount of air sucked in by the
front compressor and the release of that air. First, the converging nozzle at the front
is causing some the flow to become supersonic. Adding a diffuser can slow the flow
before it reaches the compressor. Second, to avoid flow becoming supersonic around
the capsule and in the wake, more air needs to be sucked through the front fan. A
significant amount of the air released will be directed to the air bearings to keep the
capsule levitated. However, this potentially can disrupt the capsules aerodynamics
and lead to choked airflow. The placement of the air bearings and the amount of air
flowing out will have to be carefully balanced.
25
Figure 3.8: Ansys analysis of the Hyperloop: contours of velocity magnitude (m/s).
3.7
Summary
When the pressure and blockage ratio are held constant, the aerodynamic drag is
a quadratic function of velocity. This becomes more complex when the speed of
the capsule exceeds the speed of sound, and therefore, speeds will be limited to
subsonic levels for the remainder of the thesis. At high tube pressures, the force
due to the pressure differential dominates the total aerodynamic drag value. It is
necessary to evacuate the tube to pressure levels around 100 Pa to reduce drag to
a more manageable magnitude. This will allow the capsule to obtain higher speeds
and longer coasting times. Chapter 11 will examine the effect of tube pressure on the
overall Hyperloop trip time and more general conclusions will be drawn. In addition,
for a fixed allowable drag, the lower the pressure the larger the allowable blockage
ratio. Thus, for a given capsule size the tube can be smaller, reducing material costs.
However, there is a trade-off between increasing vacuum costs and obtaining lower
pressures (Chapter 7). For a given tube environment, the smaller the blockage ratio,
the lower the drag. Again there will be a trade-off between decreasing blockage ratio
26
27
Chapter 4
Propulsion Technology for the
Hyperloop
4.1
Introduction
An advanced magnetic linear motor system will be developed to propel the Hyperloop
vehicle to travel speeds. The system will be required to maintain passenger comfort
while accelerating the capsule to 760 mph (339.75 m/s). To achieve this, the max
acceleration will be limited to 1g for passenger comfort. For roughly 90% of the
journey the capsule coasts and will not require propulsion. There will be several
major propulsion stations along the journey to get the capsule up to speed as well
as the potential to have various reboost stations every x-km to maintain capsule
speed. The tube will have a stationary element (stator) which powers the vehicle.
The capsule will contain the moving rotor element (rotor). The rotors located on the
capsule transfer momentum to the capsule via linear accelerators.
4.2
4.2.1
Linear motors are basically rotating motors that are cut and laid out flat. Instead of
producing a torque, they generate a linear force along their length. Most commonly
a linear motor operates as a Lorentz-type actuator [21]. Thus, the applied force is
28
Figure 4.1: Double-Sided Linear Induction Motor (DSLIM) schematic with magnetic
field stength inside the motor plotted [34].
The motor recommended for the Hyperloop is a Double-Sided Linear Induction
Motor (DSLIM) (Figure 4.1). A DSLIM is a LIM with a primary on both sides of the
secondary. The system benefits from having the field produced on both sides of the
secondary and as a result there is a larger flux. In addition, double sides help balance
the transverse forces acting on the rotor, keeping it aligned in the magnetic gap. These
motors have been proposed as part of the Electromagnetic Aircraft Launch System
29
4.2.2
According to the Hyperloop alpha design conception, the rotor will be attached to
the Hyperloop capsule (Figure 4.2). The alpha document listed the rotor being a
15 m long, 0.45 m tall, and 50 mm thick aluminum blade. The blade will be hallow
to reduce weight and costs, allowing current to flow in the outer 10 mm of the blade.
The distance between the rotor and stator (magnetic air gap) will be 20 mm on each
side of the rotor. The alpha document recognized that a precise control system with
electromagnetic centering would be needed to ensure the rotor safely enters, stays
within, and exits the precise magnetic air gap. The projected weight was stated to
be 1300 kg per capsule.
The stator will be mounted to the bottom of the tube. It will be 0.5 m wide,
including the magnetic air gap, 10 cm tall, and weigh 800 kg/m. The length of the
propulsion track runs 4 km. It will have a DSLIM configurationstator is laid out
symmetrically on each side of the rotor, its electrical configuration is a 3-phase, 1 slot
per pole per phase, with a variable linear pitch (0.4 m max). The number of turns
per slot also varies along the length of the stator, allowing the inverter to operate at
30
4.3
Propulsion Analysis
At this stage in the Hyperloop systems analysis it is not necessary to include all the
aspects of the Double-Sided Liner Induction Motor (DSLIM) in the propulsion models.
Instead, it is more valuable to acknowledge this technology is a viable solution for
propelling the Hyperloop capsule to speed and then focus on the general principles.
This will allow for investigation into longitudinal g-forces on passengers, propulsion
track length, propulsion times, propulsion energy, required peak power, regenerative
braking, and general cost trends.
31
4.3.1
To ensure passenger comfort the maximum longitudinal acceleration during propulsion is limited to 1g. The following analysis first uses the conditions stated in the
alpha document to determine the amount of longitudinal gs felt by riders. Then,
the limiting conditions for an acceleration of exactly 1g are determined. Results are
compiled in Table 4.1
In the alpha document the length of the propulsion track is stated to be 4 km for
each major propulsion station. To find the g force under alpha document conditions
the following relationship was developed from basic kinematic equations:
gf orce
u2e, 2 u2e, 1
/a1g
=
2dstator
(4.3.1)
where ue, 2 and ue, 1 are the capsules speed at the end of propulsion and at the
start of propulsion, respectively; dstator is the length of the propulsion track; and
a1g = 1g = 9.81 m/s2 .
The time to to accelerate under alpha document conditions was again derived
from kinematic relationships and is given below:
t =
2dstator
ue, 2 + ue, 1
(4.3.2)
Next, the limiting conditions for 1g acceleration were explored. Equation 4.3.3
calculates the minimum length of propulsion track needed to accelerate the capsule
under constant 1g acceleration conditions.
dmin =
u2e, 2 u2e, 1
2a1g
(4.3.3)
ue, 2 ue, 1
a1g
32
(4.3.4)
Table 4.1: Hyperloop propulsion: g-force, minumum track length, alpha times, and
minimum times.
4.3.2
Propulsion Segment
g force
.2292g
.5552g
.6864g
1.24g
59.65
20.93
13.61
16.88
13.67
11.62
9.34
20.96
The work and power for propulsion are important. It is assumed that the propulsion
can take place at constant acceleration, with a = 1g. The force the DSLIM needs to
supply to the vehicle is found from Newtons second law and given below:
Fprop (t) = mc g Fdrag [ue (t)]
(4.3.5)
where mc is the mass of the Hyperloop capsule and Fdrag [ue (t)] is the aerodynamic
drag (negative value) from Chapter 3 and is a function of capsule speed, ue (t).
The total work throughout the entire propulsion section is
tf
Fprop ue dt
Wprop =
t
Z itf
(4.3.6)
ti
where ti and tf are the times at the start and end of propulsion, respectively. The
total amount of work during propulsion will reveal the amount of energy required for
each propulsion section.
From elementary physics, power is defined as P = dW
= F v. The average power
dt
for propulsion will indicate the operational costs, while the peak power is a constraint
the system must be able to handle. The average power is defined by
Pavg =
Wprop
tprop
33
(4.3.7)
(4.3.8)
where ue,2 is the final speed of the Hyperloop capsule at the end of propulsion.
In terms of electrical properties, the power supplied by the DSLIM is
Pin, motor = V I
(4.3.9)
where V is the electrical voltage (potential) supplied and I is the electrical current.
The power required will dictate the electrical components used. High powers will
require heavy duty conductors and have a larger amount of heat dissipation associated
with them.
The efficiency of the DSLIM, motor , and the contributing sources of loss, especially
magnetic gap size, was discussed previously. The maximum theoretical efficiency was
found to be 0.7, but for this analysis motor is assumed to be 0.6. The degree of
efficiency demanded will directly affect the magnetic gap designed which will impact
tolerances and manufacturing costs. To propel the Hyperloop to the desired speed
the power supplied must be,
Pavg
(4.3.10)
Pin, motor =
motor
In addition, there will be an associated efficiency with the energy storage supply
transmission, trans , which will be around 0.95 in the worst case. Thus, the total
power supplied will need to be,
Pin, motor
trans
(mc g Fdrag [ue,2 ]) ue,2
=
motor trans
Pin =
(4.3.11)
Table 4.2 compiles the work and power requirements for 1g propulsion under alpha
conditions with a suspension air gap of hgap = 1.3 mm. The peak powers the system
needs to be able to handle are shown in parenthesis next to the average powers.
The results reveal peak powers on the same level as the alpha document (56 MW).
However, the average powers computed are much higher than the alpha documents
referenced 6 MW. The computed power becomes even higher when considering the
34
inefficiencies of the linear motor and the power transmission. It is possible the alpha
document considers propulsion of each section at an acceleration less that 1g and
therefore, the total time of propulsion is longer and the average power will be lower.
The alpha document also stated that solar array covering the Hyperloop is large
enough to provide an annual average of 57 MW with solar system having a 285 MW
peak power total. The energy storage of the alpha system only plans to supply the
average power (6 MW) to the propulsion, while including battery arrays at each
accelerator to deal with the peak powers. MATLAB simulations also revealed that
neither suspension gap height nor capsule frontal surface area have a large effect on
average power and total work.
Table 4.2: Hyperloop propulsion work and power requirements for 1g acceleration.
Propulsion Segment
Wprop (MJ)
134.97
327.24
404.91
732.15
4.3.3
Pavg (MW)
9.87
28.16
43.35
34.93
(19.75)
(36.57)
(50.12)
(50.12)
16.46
46.93
72.24
58.21
17.32
49.40
76.04
61.28
Regenerative Braking
There is the potential to recover some energy using regenerative braking. A regenerative braking system uses a back-to-back converter, which allows bidirectional power
flow. The electricity generated by the deceleration of the linear motor can be returned
to the grid for later use or used by auxiliary systems. Thus, some of the power required to power a capsule through the Hyperloop can be drawn from the power saved
from the braking of the previous capsule. If excess power from regenerative braking
and solar panels exceeds the input needs, there is the potential for the power to be
sold.
During regenerative braking, the load torque reverses its direction, but the operation direction remains the same. The kinetic energy from braking drives a motor
and when synchronous speed is exceeded, mechanical power is converted into electrical. The power produced from regenerative braking from capsule speed ue, 1 to ue, 2
is given by
mc (u2e, 2 u2e, 1 )
(4.3.12)
Pregen = regen
2t
where the regenerative efficiency, regen =
35
Wregen
,
Wbrake
brake energy converted into useful energy and Wbrake the amount of pure brake energy.
Typical efficiencies range between 0.6 and 0.7, and can even reach 0.85 [56].
The power and energy regenerated from braking from ue, 1 to a complete stop is
examined. The regenerative power expression becomes:
Pregen = regen (mc abrake ue, 1 )
(4.3.13)
(4.3.14)
The power regeneration and and energy recovered are plotted as functions of initial
capsule speed, ue,1 in Figure 4.3 and 4.4. It is assumed regen = 0.6 and conv = 0.95.
Regenerative Braking: Power Regeneration
35
30
25
20
15
10
5
100
150
200
250
Capsule Speed (m/s)
300
350
Figure 4.3: Regenerative braking: power as a function of initial capsule speed ue, 1 .
36
500
400
300
200
100
0
100
150
200
250
Capsule Speed (m/s)
300
350
4.3.4
68.79
206.06
166.67
372.73
Percent Recovered
50.97%
62.97%
41.16%
50.91%
Propulsion Costs
The alpha document breaks down the propulsion costs for the tube side in the following way:
37
Power Electronics
(million USD)
Energy Storage
(million USD)
$2.89
$6.99
$8.66
$3.60
$8.72
$10.78
$1.42
$3.44
$4.25
These numbers can be used to develop a capital cost for the minimum propulsion station case. In addition, the effect of adding more propulsion stations can be
evaluated using these numbers. The estimated capital cost of tube side propulsion
components is suggested to be $101.5 million for the minimum propulsion segment
case. It is interesting to note that the power electronics and energy storage compo38
nents are larger percentages of the overall propulsion costs than the alpha document
while the stator and structure are a smaller percentage. This is because the power
and energy requirements for the 1g case are very similar to the alpha case, whereas
the the propulsion track is much shorter for the 1g case than the alpha case. In
addition, not only does 1g reduce propulsion times from the alpha suggestions, the
propulsion total capital costs for 1g are less than the proposed alpha costs. This
suggests adding various small boost stations throughout the route can be achieved
for a small increase in cost. If the track path can be altered such that the track bend
radii are large, the 300 550mph and 550 760mph segments could be combined
to a single 300 760mph section. This would save propulsion costs, yet it likely the
amount of capital required to alter the track path so this is possible is extraordinary
larger than the benefits of having less propulsion stations.
To evaluate operating costs, the average cost of electricity in the Los Angles Area
for 2013 was found to be $0.214 per kW-hr (Bureau of Labor Statistics, 2014 [6]).
Assuming the Hyperloop is operational for 12 hours a day, year-round with an average capsule departure of 2 minutes, then annual cost of operating the minimum 6
propulsion stations is $6.77 million USD without regenerative braking and $3.32 million USD with regenerative braking. These operational costs can be reduced by using
a solar array. Assuming the solar array spans the length of the tube and generates an
average of 57 MW annually, there is more than enough energy for propulsion. This
additional energy could be sold for approximately $100.08 103.53 million USD per
year. A little over two years of operation could pay for the capital costs of the solar
array and battery infrastructure ($260 million USD). Because more than propulsion
will need energy input (vacuum pumps for example) this number will be much lower,
but still significant enough to sell and ultimately keep ticket prices low.
39
Chapter 5
Air Bearing Suspension System
5.1
Introduction
5.2
Alpha Design
The alpha design parameters will direct the analysis of the flow behavior in the
suspension gap as well as the air bearing suspension systems feasibility and control.
The alpha document design has air bearing suspension skis floating on a pressurized cushion of air 0.50 mm to 1.30 mm off the tubes bottom (Figure 5.1). This
40
Figure 5.1: Alpha design schematic of air bearing skis that support the capsule [34].
distance is defined as the suspension gap height (hgap ). Due to the compression of air
in the small gap and heat transfer, the temperature will increase (Tgap = 398.15 K)
and effect the flow. The alpha team concluded only 9.4 kPa is needed to support the
passenger capsule. The flow field in the gap will exhibit a highly non-linear reaction
when the gap height between a ski and the tube is reduced. A large restoring pressure
results, pushing the ski away from the wall and back to its nominal ride height. The
skis will be constructed from iconel, a trusted alloy of SpaceX that is stiff and can
withstand high pressure and heat. To eradicate any discomfort to passengers, each
ski will be integrated into an independent mechanical suspension. The design also
has the possibility of adding deployable wheels for speeds under 100 mph to ease
movement and increase overall system safety.
The suspension parameters used during numerical analysis can be found in Table
5.1.
Alpha Suspension Parameter
hgap, min (mm)
hgap, max (mm)
Wski (m)
Lski (m)
Nski
(deg)
pgap (kPa)
Tgap (K)
gap (N-s/m2 )
Value
0.5
1.3
0.9
1.5
28
0.5
9.3
398.15
2.3185E05
41
5.3
5.3.1
The average gap temperature can be estimated from an adiabatic temperature loss
computation,
T2 = T1
= T1
V1
V2
1
h1
h2
1
(5.3.1)
(5.3.2)
where h1 is the distance the front of the air bearing ski is off the bottom of the tube,
h1 = hgap + Lski sin()
where is the skis angle of attack. h2 is the distance the back of the air bearing ski
is off the bottom of the tube,
h2 = hgap
T1 is the temperature at the ski front (T1 = Ttube ) and T2 is the temperature at the
ski rear. The average temperature in the gap is thus,
Tgap, avg
1
= Ttube
2
1
(5.3.3)
Using Ttube = 293.15 K and the maximum gap, hgap, max , the adiabatic temperature
in the gap would be Tgap, avg = 383.45 K.
In reality, there will be heat transfers due to friction so the adiabatic conditions
will not hold and the average calculation will not be appropriate. Correct analysis of
the temperature will consider the viscous dissipation which depends on the velocity
gradient in the gap. This will yield a value closer to what the alpha document
presented. Thus, it is assumed Tgap = 398.15 K.
5.3.2
Due to the small gap size between the tube wall and the suspension ski, the value
of the Knudsen number needs to be computed in order to determine whether the
42
hgap
kB Tgap
2 2 pgap hgap
(5.3.4)
where kB is the Boltzmann constant and is the particle hard shell mean diameter
( 3.7 1010 m for air). For the minimum gap height, the largest Knudsen value
observed would be 0.002. Therefore, it is a valid assumption to use continuum fluid
mechanics.
5.3.3
Couette Flow
Chapter 3 identified one of the contributing factors to the aerodynamic drag was air
friction. Couette flow was used to describe the flow between the tube wall and the
capsule. The tube wall is treated as the fixed surface and the capsules suspension
bearing ski can be represented by a moving plate at velocity ue . Due to the no-slip
condition, there can be no relative motion between the plate and the fluid. The upper
plate exerts a shear stress on the fluid causing the fluid to move. For, incompressible
flow the velocity profile was found to be,
u = ue
y
hgap
(5.3.5)
The fluid flow profiles of the three main Hyperloop speeds are plotted for the minimum
suspension gap height (0.5mm) in Figure 5.2 and for the maximum suspension gap
height (1.3mm) in Figure 5.3. These figures reveal that Couette flow exhibits a linear
profile.
43
300
Velocity (m/s)
250
200
150
100
umax
50
umid
u
min
0
0
0.1
0.2
0.3
0.4
Distance from Bottom of Tube (mm)
0.5
Figure 5.2: Couette flow analysis for gap = 0.5 mm and ue, max = 339.75 m/s, ue, mid =
248.11 m/s, and ue, low = 134.11 m/s.
Gap Velocity Profile: Couette Flow (1.3mm)
350
300
Velocity (m/s)
250
200
150
100
u
50
max
umid
u
min
0
0
0.2
0.4
0.6
0.8
1
Distance from Bottom of Tube (mm)
1.2
Figure 5.3: Couette flow analysis for gap = 1.3 mm and ue, max = 339.75 m/s, ue, mid =
248.11 m/s, and ue, low = 134.11 m/s.
44
Newtons viscous effect states that the shear stress in the fluid is proportional to
the rate of change of velocity with respect to y. Thus, the shear stress is given by
=
u
y
(5.3.6)
ue
D
(5.3.7)
Fluids exhibiting this characteristic are known as Newtonian Fluids. The max Couette shear stresses for the three main capsule speeds are compiled in Table 5.2.
5.3.4
Lubrication Theory
dp
d
Fx = p dy dx p +
dx dy dz dx dz + +
dy dx dz = 0 (5.3.8)
dx
y
This reduces to
dp
=
dx
y
46
(5.3.9)
(5.3.10)
1 dp 2
y + C1 y + C2
2 dx
(5.3.11)
The constant C1 and C2 can be found by applying the two boundary equations
At y = 0, u = 0
At y = h, u = ue
The final equation becomes
u=
1 dp 2
ue
(y hy) + y
2 dx
h
(5.3.12)
This velocity in the gap now depends on both the coordinate y and the pressure
dp
. The velocity profile is a superposition of a linear profile and a parabolic.
gradient dx
This flow is often called Couette-Poiseuille Flow, which superimposes plate flow with
dp
pipe flow. If dx
is equal to 0, then the linear Couette velocity profile will be observed.
However, there will be a pressure gradient in the Hyperloop suspension air gap and
dp
a more parabolic profile will be observed. Setting dx
= 9300, the resulting velocity
profiles can be seen in Figures 5.5 and 5.6. The parabolic nature of the velocity is
very evident in the plot for the hgap, max .
47
300
Velocity (m/s)
250
200
150
100
umax
50
umid
u
low
0
0
0.1
0.2
0.3
0.4
Distance From Bottom of Tube (mm)
0.5
Figure 5.5: Journal bearing analysis for gap = 0.5 mm and ue, max = 339.75 m/s,
ue, mid = 248.11 m/s, and ue, low = 134.11 m/s.
Gap Velocity Profile: Journal Bearing Analysis (1.3mm)
350
300
Velocity (m/s)
250
200
150
100
umax
50
umid
u
low
0
0
0.2
0.4
0.6
0.8
1
Distance From Bottom of Tube (mm)
1.2
Figure 5.6: Journal bearing analysis for gap = 1.3 mm and ue, max = 339.75 m/s,
ue, mid = 248.11 m/s, and ue, low = 134.11 m/s.
48
5.3.5
Mass and momentum conversation can be applied to a control volume around the air
suspension ski with compressed air input enabled (Figure 5.7).
i
1 h 2
(u1 h1 u22 h2 ) + p1 h1 + Lski cos()
h2 2
(5.3.14)
where is the shear stress; Lski is the length of suspension ski; and is the angle of
attack. Momentum conservation in the y direction provides a way to relate the lift
force (FL ) to the pressure and mass flow rate from suspension system.
FL = FW, ski Lski cos() (ps + p)
49
(5.3.15)
5.3.6
Control Overview
The tube wall will not be a constant, smooth surface throughout the Hyperloop
capsules journey. The capsule will need to adapt to the changing tube landscape.
A precision sensor will be needed to scan for upcoming step size changes in the
tube bottom walls contour. Investigating the control will reveal time constants,
sensing distances, and maximum allowable tube changes after a given initial change.
The response of the air bearing ski will be approximated by modeling a pistons
movement. The piston chamber is the region between the tubes bottom wall and
the air bearing ski with a width of Lski cos() and piston (ski) position to the wall
50
1 2
2pgap + ue
2
(5.3.17)
(5.3.18)
4
3.5
3
2.5
2
1.5
0 mm
1
0.5
0
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
Time (s)
Figure 5.8: Air bearing ski position as a function of time since initial bump: uncontrolled response for gap height = 1.3mm
The restoring force of the bearing ski is given by p Aski and a differential
equation can be solved for a given tube contour change. The zero position indicates
when the rear of the ski comes into contact with tube bottom. For uncontrolled
response, the oscillatory position can be seen in Figure 5.8. The larger the bump
encountered, the more the ski overshoots the nominal gap height (1.3mm). For all
bump heights, this overshoot of the nominal is larger than the subsequent undershoot.
The period of oscillation also increases as the initial bump size increases. A suspension
51
control mechanism is needed to prevent the oscillations and bring the ski to rest at the
nominal gap height. If a sophisticated control system is absent, the capsule may be at
risk of reaching harmonic resonance in the tube. Similar to pilot-induced oscillations
(PIOs) [43], sustained or uncontrollable oscillations can result (due to reduced phase
margin) and lead to capsule failure.
Max Permissible 2
nd
1.4
0mm
Permissible 2
nd
1.2
1
Increasing Initial
Bump Height
0.8
0.6
0.4
0.2
1.3mm
0
0
0.002
0.004
0.006
Time (s)
0.008
0.01
0.012
Figure 5.9: Maximum incoming tube bump height as a function of time since initial
bump was encountered for gap = 1.3 mm.
How fast the skis position returns to nominal height is very important as it will
reveal how far in advance an impermissible bump/tube change must be detected.
Figures 5.9 and 5.10 show the height of a permissible second bump as functions of
time and distance traveled since the initial bump, respectively. These plots focus on
the position response up until it reaches nominal height. To simplify the analysis
it will be assumed a control mechanism will be able to keep the ski close to the
nominal height and return to equilibrium in a short amount of time after reaching
the nominal height. Larger initial bumps allow smaller permissible second bumps at
first, but return to nominal height faster. This effect can be observed in the figures.
The necessary distance to scan in front of the vehicle will depend on how fast the
measuring device can sample the ground and how fast the an actuator can respond.
52
Max Permissible 2
nd
1.4
0mm
Permissible 2
nd
1.2
1
Increasing Initial
Bump Height
0.8
0.6
0.4
0.2
1.3mm
0
0
0.5
1.5
2
2.5
Distance (m)
3.5
Figure 5.10: Maximum incoming tube bump height as a function of distance traveled
since initial bump was encountered for gap = 1.3 mm.
In addition, since the Hyperloop capsule only floats on a thin film of air (0.5
1.3mm) the effect of human movement on the displacement of the capsules center
mass is very important. Large movements can help induce harmonic resonance and
lead to an effect similar to PIOs, causing failure. Figure 5.11 shows the displacement
of the capsules center of mass downward for corresponding vertical movement of a
passenger of mass 100 kg. The horizontal red lines represent the boundaries for the
air gap heights. It can be seen even a small human movement upward can result in
the capsule center of mass moving downward by a larger amount than either the min
or max gap heights. Although the large restoring pressures may eventually return
the capsule to nominal cruising height, a very robust control system with precise
actuators and sensors and a fast time constant still needs to be developed in order to
avoid the risk of harmonic resonance.
The center of mass equation in (5.3.19) was used to develop figure 5.11.
ydisp =
mh
ymove
mc
(5.3.19)
Chapter 6 is a case study evaluating the Hyperloop air bearing suspension system
using the quarter car suspension model and applying H control synthesis.
53
1.6
Max Gap Height
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0
50
100
150
Human Movement (mm)
200
250
Figure 5.11: Vertical displacement of capsule center of mass due to human movement.
5.4
54
Figure 5.12: Capital construction costs for the proposed California-Nevada Interstate
Maglev Project (CNIMP)[10].
The Revised 2012 Business Plan for the California High Speed Rail (CHSR) proposed between Anaheim and San Francisco paints a slightly different picture. Out of
the proposed $68.4 billion expenditure capital cost, only $2.046 billion is estimated
for the trackwork [8]. Again this document does not specify what comprises this
category, and it may be assumed this trackwork term encompasses all the maglev
components on the track side. Thus, the track side of maglev is only about 3% of the
entire budget. For a proposal that appears to be more thought out, these numbers
may be more accurate. Returning to the issue of category ambiguity, the report also
lists capital costs for structures and civil, and these are 13.5 times that of trackwork
($22.4 billion and and $5.35 billion, respectively). It is unclear whether these categories contain any essential parts specific to the maglev technology or if they are
instead just overall structure costs that will be comparable to similar aspects like the
Hyperloop tube and pylons.
A 1997 Journal of Transportation Engineering article assessed the maglev guideway cost of a potential route from Baltimore (BAL) to Newark (NWK) (264 km). The
assessment considered four potential system concept designs and the cost of guideway
construction cost per km ranged from $7.6 million to $21.1 million [42]. Accounting
56
for inflation and extrapolating to the Hyperloop route yields a cost range between
$6.37 and $17.7 billion. This cost categorization does offer some more details. It mentions the estimates do not include right-of-way, propulsion, station, and other fixed
facilities. The data is meant to purely estimate the construction of the guideway
structure. About 20% of the estimated costs are due to labor.
The proposed guideway infrastructure costs for German Intercity-Express (ICE)
and Transrapid routes range from $30.76 to $43.43 million (USD) per km (accounting
for inflation) [51]. Based on the other costs presented, the guideway infrastructure
category in this case is expected to encompass much more than maglev technology
related components. Under the assumption that 50% of this category is related to
maglev components, it would correspond to $8.77 to $12.38 billion for the Hyperloop
route. These findings are compiled in Table 5.3.
Table 5.3: Maglev guideway cost assessment extrapolated to Hyperloop route
System
NA
NA
$1.82
$6.37
$8.77
$6.0
$7.6
$2.05
$17.70
$12.38
Maglev technology costs in high speed rail systems were found to vary significantly
due to the ambiguity of cost categories. Some of these costs from the literature, in
addition to raw maglev capital costs, may include a lot of the basic infrastructure
that is needed for a Hyperloop implementing an air suspension system. It is hard to
distinguish what percentage of these values would be an added cost to a Hyperloop
featuring an air suspension system instead of maglev suspension. It is imperative
that accurate cost estimates are available to access the financial viability of maglev
technology in the Hyperloop. More research is needed to determine if this technology
should be given more serious thought. An advantage of maglev is the developed stage
of the concept and its successful use in many operational systems. The air suspension
system will need to be prototyped on a large scale and this process is likely to run into
many challenges, adding a risk factor. The tradeoffs between using a proven system
at a potentially much higher cost and spending the time/research to implement a new
suspension solution on a large scale need to be considered.
57
Chapter 6
In-Depth Analysis of the Quarter
Car Suspension System using H
Control Synthesis
6.1
Introduction
A joint effort for this thesis and a final project for MAE 434: Modern Control was to
analyze a design a control system for the Hyperloops air cushion suspension system.
The derivation of the system dynamics led to a very complex system that would make
the design and analysis process challenging. The literature was first consulted in order
to find a similar environment with a simplified system analogous to the Hyperloops.
Research on air suspension systems (Quaglia & Sorli, 2001 [45]), air cushion vehicles
(Shuo et al., 2006 [50]), hovercrafts (Sira-Ramirez & Ibanez, 2000 [52]) and maglev
systems (Cai & Chen, 1996 [7]) revealed models that either did not fit with the
Hyperloops or remained too large of scope to effectively navigate the system. While
considering alternative models, the quarter car suspension system appeared quite
frequently. Although there were some key differences between this model and that of
the Hyperloops suspension system, the model is simple to analyze and there is the
opportunity to gain basic understanding of the suspension concept and later translate
the findings to the Hyperloops system. Stable operation and acceptable ride comfort
requires vehicle motion control in both cases. Passive systems only store or dissipate
energy associated with vibration. This is insufficient to improve the performance
58
requirements and be robust to road roughness and other transient disturbing forces.
Therefore, an active suspension system is needed. For the quarter car suspension
system H techniques were first used to design a controller for the nominal model.
Then, synthesis was explored to design a controller that was robust to uncertainty
in the model.
6.2
The quarter car suspension system can be modeled as a one-dimensional vehicle with
two degrees of freedom (Figure 6.1). The mass of the capsules body (chassis) is
represented by mb and the mass of the wheel assembly (air bearing ski) is mw . Between
the capsule body and the wheel assembly there is a passive spring, ks , and a shock
absorber (damper), bs . The compressibility of the pneumatic tire (the air cushion) can
be modeled by kt . The active component of the suspension system is an applied force,
fs , applied between the body and the wheel assembly, while the compressor force,
fcomp , is applied between the wheel assembly and the road (tube bottom wall). In the
literature there were other models found for this system. The main differences were
an additional damper in the wheel assembly and moving the location of the applied
force to between the road and the wheel assembly. To simplify this system in order to
observe the overall control effect on road handling and comfort, the compressor force
will not be considered as an actuated force. Thus, the only actuated force input will
be, fs .
59
(6.2.1)
mb xb + bs (x b x w ) + ks (xb xw ) = fs
(6.2.2)
1
[ks (x1 x3 ) + bs (x2 x4 ) fs ]
mb
x 3 = x4
1
[ks (x1 x3 ) + bs (x2 x4 ) kt (x3 r) fs ]
x 4 =
mw
= Ax(t) + Bu(t)
y(t) = Cx(t) + Du(t)
where,
ks
m
b
A=
0
ks
mw
C= 1
ks
mb
bs
mb
ks
mb
bs
mb
bs
mw
ks kt
mw
0
0
0
1
bs
mb
ks
mb
,
1
bs
mw
0 , and
bs
mb
0
B=
0
kt
mw
D = 0
0
fs
mb
,
0
fs
mw
0
0
fs fcomp
mb
The model inputs are the road disturbance, r, and the actuator force, fs .
The model outputs are the car body travel, xb , the suspension deflection, sd = xb xw ,
60
System Parameter
Value
mb (kg)
mw (kg)
bs (N/m/s)
ks (N/m)
kt (N/m)
425.0
110.7
100
16000
190000
6.3
A hydraulic actuator connected between the body mass and wheel assembly mass
is used for active suspension control. There are five main components: the electro
hydraulic powered spool valve, piston-cylinder, hydraulic pump, reservoir, and piping
system (Figure 6.2). The actuator considered has a max displacement of 0.05 meters
[30]. The nominal model is an approximation of the physical actuator dynamics and
can be represented by the transfer function,
ActN ominal =
1
+1
1
s
60
1
s + 1
where K is the servo valve static flow gain at zero load pressure drop and is the
apparent servo valve time constant [57]. If a wider frequency range is desired, a second
order response can be used.
Model variations in the actuator are considered for the robustness analysis.
61
6.4
Preliminary Analysis
62
Gain from Road dist (r) and Actuator Force (fs) to B Accel (ab) and Suspension Travel (sd)
100
To: ab
50
0
50
Magnitude (dB)
100
150
50
To: sd
0
50
100
150
200
0
10
10
Frequency (rad/s)
10
Figure 6.3: Bode Plot Showing the Tire Hop and Rattlespace Frequencies
63
6.5
H Control Synthesis
The goal of H methods is to synthesize a controller that minimizes a single cost function. The optimization considers factors such as passenger comfort, road handling,
quality of sensor measurements for feedback, and possible limits on available control
force. Weighted parameters are used to model external disturbances and emphasize
design objectives. The optimization takes place over the Banach space H , which
consists of all complex-valued functions of a complex variable which are analytic and
bounded in the open right-half complex plane defined by Re(s) > 0. The H norm is
the maximum singular value over this space (peak gain). The synthesis can be used
to minimize the closed-loop impact of perturbations.
64
limits the control bandwidth is used for Wact . The weighting functions Wd2 and Wd3
were adjusted to model broadband senor noise.
Target transfer functions for the closed-loop response of the system were established to represent goals of passenger comfort and adequate road handling. These
were used to help determine the weighting functions (Wsd and Wab are reciprocals of
the comfort and handling targets). The target transfer functions are plotted vs. the
open-loop response for the road disturbance to the suspension deflection and body
acceleration in Figure 6.5.
Response to Road Disturbance
From: d1
0
To: sd
20
40
Magnitude (dB)
60
Openloop
80
Closedloop Target
60
To: ab
40
20
0
20
40
0
10
10
10
10
Frequency (rad/s)
65
Condition
H norm
= 0.1 (Comfort)
= 0.9 (Handling)
= 0.5 (Balance)
0.632
0.766
1.066
66
To: xb
50
10
0
To: sd
Magnitude (dB)
100
20
10
20
Openloop
30
Comfort
40
60
Balanced
Handling
50
To: ab
40
30
20
10
0
0
10
10
Frequency (rad/s)
10
Figure 6.6: Closed-loop for H synthesized controllers for each blending parameter,
, compared to passive suspension system (open-loop)
67
The effect of the blending parameter, , for the design of the H controller is
more easily observed in the time response. Figure 6.7 shows simulations for a road
(tube surface) disturbance signal (green curve) characterized by,
(
r(t) =
The tradeoffs of comfort and handling are very evident. For example, the body
acceleration is the smallest for the controller emphasizing comfort (red curve) and
is the largest for the controller emphasizing handling (black curve). The suspension
deflection is minimized for the handling case and is worse for the comfort case. However, the controller that balances the goals actually provides a very good compromise
for all responses (pink curve). This would be the recommended controller to use.
Body Travel
Body Acceleration
0.1
0.8
0.05
ab (m/s2)
xb (mm)
0.6
0.4
0.2
0
0.05
0.2
0.4
0.2
0.4
0.6
0.8
0.1
0.2
Suspension Deflection
1
0.4
0.6
0.8
Control Force
x 10
Openloop
Comfort
Balanced
Suspension
Road Disturbance
1.5
1
0.5
fs (N)
sd (mm)
0.5
0.5
0
0.5
1
1.5
1.5
0.2
0.4
0.6
Time (s)
0.8
0.2
0.4
0.6
Time (s)
0.8
Figure 6.7: Time Response for Road Disturbance Signal: Effect of Blending Parameter,
68
6.6
Robust Synthesis
The H controller was constructed for only the nominal actuator model, which is
an approximation of the true actuator. The design needs some further work to ensure performance is maintained even with model error and uncertainty. The robust
controller was synthesized using D-K iteration for the model with actuator uncertainty. The MATLAB command dksyn(P,NMEAS,NCONT) synthesizes a controller
K for the open-loop plant model P via the D-K iteration approach to -synthesis. P
is a state-space model with uncertainty incorporated and it is assumed that the last
NMEAS outputs and NCONT inputs of P are the measurement and control channels,
respectively. The balanced model case was examined. The closed-loop response for
the robust design is compared to the open-loop (Figure 6.8). The responses are very
similar to the balanced H -only controller in the absence of uncertainties.
Body Travel
Body Acceleration
0.1
0.8
0.05
ab (m/s2)
xb (mm)
0.6
0.4
0.2
0
0.05
0.2
0.4
0.2
0.4
0.6
0.8
0.1
0.2
Suspension Deflection
1
0.4
0.6
0.8
Control Force
x 10
0.5
0
fs (N)
sd (mm)
0.5
0
2
4
1
1.5
Openloop
Robust Design
6
0
0.2
0.4
0.6
Time (s)
0.8
0.2
0.4
0.6
Time (s)
0.8
Figure 6.8: Time Response for Road Disturbance Signal: Robust Synthesized Controller vs. Open-Loop.
69
10
x 10
To: xb
Amplitude
5 x 103
1
To: sd
2
0.1
To: ab
0.05
0
0.05
0.1
0.1
0.2
0.3
0.4
0.5
Time (seconds)
0.6
0.7
0.8
0.9
Figure 6.9: Time Response for Road Disturbance Signal: H Nominal Controller
(red) and Model Uncertainty (blue)
70
To: xb
10
x 10
To: sd
Amplitude
5 x 103
1
2
0.1
To: ab
0.05
0
0.05
0.1
0.1
0.2
0.3
0.4
0.5
0.6
Time (seconds)
0.7
0.8
0.9
Figure 6.10: Time Response for Road Disturbance Signal: Synthesized Robust
Controller (red) and Model Uncertainty (blue)
71
6.7
A paper by Li, Jing, and Karimi (2013) investigated output-feedback-based H control for the quarter car suspension system with control delay. They showed that
adding a control delay improved performance, especially achieving less body acceleration and further improvement for ride comfort [27]. Actuator delay also proved
more effective in the presence of road disturbances. Therefore, they argue actuator
delay is crucial in vehicle suspension. However, this may not be applicable to the
Hyperloop. This system will need a very quick response and a control system with a
fast time constant. A time delay may not be possible to implement.
LQG/LTR analysis was also emphasized in the literature for the quarter car suspension system (ElMadany & Abduljabbar, 1999 [16]), and similarly with the maglev
system (Cai & Chen, 1996 [7]). This analysis was also implemented and results found
were similar to what was seen in previous papers. They showed LQG/LTR was effective for this system. However, these results were not as interesting as the H design
so they were not included in this report.
6.8
Summary
H controller synthesis with robustness design effectively balanced passenger comfort and handling for the quarter car model of the Hyperloop and was robust to model
error and uncertainty. This control method was only applied to a simplified version
of the Hyperloop suspension so it would be interesting to apply this control design
to a more developed Hyperloop system model and compare the results found. In
addition, the effect of control delay should be investigated first for the quarter car
system and then considered for the Hyperloops system. However, when considering
the results of H synthesis it should be noted that the resulting controller is only
optimal for the cost function created. It may not be the optimal for performance
measures typically used to evaluate controllers such as settling time, overshoot, or
oscillatory behavior. There are many control methods and they may or may not end
up with the same result. It is important to consider other methods and the potential
benefits and drawbacks to using a different controller. In essence, effective control
design is truly an art.
72
Chapter 7
Vacuum Pump Technology for the
Hyperloop
7.1
Introduction
Vacuum pumps are essential to evacuate air from the Hyperloop tube before operation
and to maintain tube pressure throughout the systems life. Vacuum pump stations
will be built at regular intervals along the tubes length. Airlock chambers will be
needed to transfer capsules from the open atmosphere to the vacuum and vice versa.
This chapter analyzes the factors that affect tube pump-down times (vacuumizing
the tube). The vacuumizing time affects the entire operation of the whole Hyperloop
system. The factors that impact pumping times include orifice conductance, diameter
of the tube, length of the tube, space between pumping stations, effective vacuumizing
speed, and the target vacuum pressure. Based on desired properties available vacuum
technologies were investigated and device suggestions were made.
7.2
7.2.1
An orifice links the vacuumizing pipe between the pump station and the Hyperloop tube. Flow resistance occurs as a result of external friction (gas molecules/wall
7
73
surface) and internal friction (gas molecule/gas molecule, i.e. viscosity). The conductance will affect volume flow rate (pumping speed) of the vacuum. The orifice
conductance can be calculated from the following formula,
r
C0 = 3.36
T
A0
M
(7.2.1)
where T is the ambient temperature (K), M is the molar mass of the gass (g/mol),
and A0 is the orifice area (cm2 ).
7.2.2
T
M
(7.2.2)
where D is the tube diameter (cm), L is the tube section length (cm), and the other
parameters are defined above.
Since the tube and the vacuumizing pipe are connected in series the equivalent conductance can be determined from a formula analogous to springs in series,
1
1
1
=
+
C
C0 C1
(7.2.3)
From vacuum theory the available vacuumizing speed can be determined from the
following formula:
1
1
1
=
+
(7.2.4)
Se
S0 C
where Se is the available vacuumizing speed (L/s), S0 is the nominal vacuumizing
speed (L/s), and C is the system conductance (L/s).
7.2.3
Vacuumizing Time
For a target tube pressure, p, tube volume, V , and available vacuumizing speed, Se ,
the pump down time can be calculated from
t=
V
p0
ln
Se
p
74
(7.2.5)
Vacuumizing time is thus directly proportional to tube section volume and inversely
proportional to available vacuumizing speed of the selected pump (see schematic in
Figure 7.1).
7.3
7.3.1
The orifice conductance (C0 ) was computed using a typical orifice diameter (D0 ) and
other system parameters displayed in Table 7.1. The orifice conductance was found
to be C0 = 3561 L/s for this configuration.
Vacuum Parameter
D0 (cm)
A0 (cm2 )
T (K)
M (g/mol)
C0 (L/s)
Value
20.0
314.16
293.15
29
3561
vacuuminzing speed was determined. The available vacuumizing speed was plotted
using MATLAB as a function of tube diameter for various station intervals in Figure
7.2.
Available Vacuum Speed vs. Tube Diameter
1200
500 m
1000
800
600
5000 m
400
200
0
2
4
Tube Diameter (m)
Figure 7.2: Available vacuumizing speed as a function of tube diameter for pump
station interval distances from 500 m to 5000 m
The pump-down time was computed using Equation 7.2.5 for pump stations every
2 km; tube diameters 2.23 m (alpha), 3 m, 4 m, and 5 m; and available vacuumizing
speeds, Se , 200 L/s, 500 L/s, and 1000 L/s. The exponential dependence of pumpdown time on target pressure can be observed in Figure 7.3. Lower target pressures
will take longer to achieve and require more energy input to reach these levels. Since
vacuumizing time directly depends on the square of the tube diameter, reducing the
tube section diameter as far as possible will help shorten vacuumizing times and
energy costs. Figure 7.4 takes a closer look at the pressures in the range of an
operational Hyperloop (0-200 Pa). Starting at 200 Pa, the time only slowly increases
until the target pressure is around 30 Pa. Upon reaching this point, the time begins to
shoot up. Combined with the aerodynamic results seen in Chapter 3 that suggested
only minimal drag reduction can be achieved by reducing the pressure past 40 Pa
and the increased pumping times beyond this point, it is not recommended the tube
pressure be less than 30-40 Pa for this vacuum pump configuration. Other vacuum
configurations can be explored, but it becomes increasingly difficult to get higher
vacuumizing times as the size and the complexities of these vacuum pumps increase.
76
77
1000
2000
3000
4000
5000
6000
7000
8000
9000
2000
4000
6000
Tube Pressure (Pa)
D = 2.23m
D = 5m
Se = 200 L/s
8000
10000
1000
1500
2000
2500
3000
3500
4000
2000
4000
6000
Tube Pressure (Pa)
D = 2.23m
D = 5m
Se = 500 L/s
8000
10000
200
400
600
800
1000
1200
1400
1600
1800
2000
2000
4000
6000
Tube Pressure (Pa)
D = 2.23m
D = 5m
Se = 1000 L/s
8000
10000
Figure 7.3: Vacuum pump-down time as a function of tube pressure for D = 2.23, 3, 4, and 5 m; Se = 200, 500, and 1000 L/s
10000
Se = 1000 L/s
4000
3500
3000
2500
D=5m
2000
1500
1000
D = 2.23 m
500
0
50
100
Tube Pressure (Pa)
150
200
Figure 7.4: Vacuum pump-down time for tube pressures P = 0 - 200 Pa; tube diameters D = 2.23, 3, 4, and 5 m; available vacuumizing time Se = 1000 L/s
78
7.3.2
14
12
10
8
6
Increasing S0
4
2
0
0
50
100
Airlock Pressure (Pa)
150
200
Figure 7.5: Airlock pump-down time as a function of target pressure for Dtube = 2.23
m and alpha document capsule size; Nominal vacuumizing speed ranges linearly from
1000 L/s to 12,000 L/s.
shape seen in the previous section. Furthermore, it can be seen that as nominal
vacuumizing speed increases the curves begins to converge. This indicates it will be
harder to achieve even faster pump-down times. The curves converge to around 40 s
to 80 s airlock vacuumizing times for tube pressures of interest and S0 ranging 8,000
L/s to 12,000 L/s. To indicate the difficulty of decreasing this time, to decrease the
time in half the nominal vacuumizing speed, S0 , would need to be slightly more than
79
double (holding everything else constant). For example, to decrease pump-down time
from 40 seconds to 20 seconds given a desired airlock pressure of 100 Pa, S0 increases
from 11,931 L/s to 23,954 L/s. This is a factor of 2.01 increase. Obviously, faster
pump-down times will be desired in the airlock to make the trip length including
loading/unloading minimal, but the costs of achieving faster and faster pump-down
times begin to far exceed the benefits of faster times.
7.3.3
Power Requirements
The power requirements of the vacuum pump stations will contribute to overall system
energy requirements. The average power requirements can be estimated using the
following equation:
S0 P
P =
(7.3.1)
mech
A typical motor used for vacuum pumps will have a mechanical efficiency of mech =
0.85. Using the configuration examined in the previous subsections with S0 = 1800
L/s and a target pressure of 100 Pa, Equation 7.3.1 yields an input power of 214.36
kW per pump for the initial pump down from atmospheric pressure.
7.4
When high flow rates (vacuumizing speeds) are required, a rotary vane backing pump
can be combined with a Roots pump.
A rotary vane vacuum pump (Figure 7.6) is an oil-sealed rotary displacement
pump. The electrically installed rotor (2) and the valves (3), which move radially
under spring force, divide the working chamber into two separate spaces with variable
volumes. As the rotor turns, gas flows into the enlarging suction chamber until it is
sealed off by the second vane. The enclosed gas is compressed until the outlet valves
opens against atmospheric pressure. Rotary vane pumps can be built in single- and
two-stage versions.
Roots pumps (blowers) are a type of dry-running rotary displacement vacuum
pump (Figure 7.7). They feature two synchronously counter-rotating rotors (4) in a
figure-eight configuration separated by a narrow gap. One shaft is driven by a motor
(1) while the other is synchronized by a pair of gears (6). Very high rotary speeds
80
81
Figure 7.8: Pfieffer CombiLine 1902 - Roots pump with single-stage rotary vane
pump: Pumping speeds as function of pressure [40]
with a Roots pump. The pumping speed characteristics can be observed in Figure
7.8 and it is evident the pumping speed is lower when the pump initially begins with
atmospheric conditions. The available vacuumizing speed then increases once the
tube pressure hits 100 hPa (1000 Pa) as the Roots pump has fully kicked in and
provides extra boost to pumping speed. The maximum pumping speed is 2200 m3 /h
(611 L/s) and the ultimate pressure ranges from 0.3 Pa to 0.08 Pa. The power spec
is around 22 kW with 230 V or 460 V options for the three phase motor [40]. Using
Equation 7.3.1 with S0 = 2200 m3 /hr and target pressure of 100 Pa yields a power
of 72.8 kW, more than three times the observed value. Thus, Equation 7.3.1 may
not be the best for estimating the power required for a Roots pump backed with a
single-stage rotary vane pump. In addition, the power input to the drive motor will
change as intake pressure reduces. While the data was not available for this unit, the
behavior will be similar to the motor supply to Oerlikon Leybold Vacuums rotary
plunger pump in Figure 7.9 despite the pumping speed being much less (Se = 60
m3 /hr)[36].
Although the Pfieffer CombiLine WU 1902 has a slightly lower pumping speeds
than desired for the Hyperloop, especially for the airlock vacuumizing, it is the largest
commercially sold of its type with a price tag. A representative from Pfieffer discussed
larger pumps with faster pumping speeds can be designed and sold, but prices will
83
Figure 7.9: Rotary plunger pump (Se = 60 m3 /hr) motor power as a function of
intake pressure an operating temperature [36]. Other pumps will have similar power
behavior.
vary for custom projects. The Combiline WU 1902 will cost $23,900 for a single unit
and $17,000 for mass order. The vacuum company recently sold a single unit of a
Roots pump backed with a single-stage rotary vane pump with pumping capabilities
of 27,000 m3 /hr (7500 L/s) for $250,000. While this high magnitude of pumping speed
may not be needed for initial pump-down, this pumping station would be more ideal
for the Hyperloop airlock interface. The maximum nominal vacuumizing speed seen
in these type of pumps is 12,000 L/s [67]. The ultimate pressure of Pfieffer CombiLine
WU 1902 pump station is lower than needed for Hyperloop operation. In part, this
is necessary because a degree of oversize is required for initial pump down. Yet, as
indicated earlier, when a pumping station has a lower ultimate pressure than desired a
control mechanism will be necessary. A further factor to consider is that these pumps
cannot be left outside and exposed to rain, snow, etc. The temperature limits for a
station like the Pfieffer CombiLine is 12 40 C. This is a reasonable temperature
range for an area like Southern California, but further studies are required to establish
84
what happens when it gets too hot or cold, and if an air conditioned room with heating
capabilities for the winter is necessary.
7.5
Summary
This chapter revealed that vacuum pump-down time goes as the square of the diameter, and thus the smaller the tube the faster the target pressure can be achieved. In
addition, higher vacuumizing speeds shorten the pump-down time, but vacuum pump
stations become more complex and expensive to build/operate as higher pumping
speeds are demanded. The interval distance between pumping stations is very important as pump-down time is directly proportional to this distance. The shorter the
interval, the faster the times. A balance needs to be achieved between the amount
of pumping station infrastructure, pumping times, and effect of station servicing. A
suggested interval distance is 2 km between pumping stations. The desired degree of
vacuum in the Hyperloop is the most important factor as vacuumizing time increases
exponentially as target pressure decreases. In combination with the results of aerodynamic drag and pressure it is suggested tube pressure not be reduced below 40 Pa.
For the alpha document conditions with a target tube pressure of 100 Pa, pumping
stations every 2 km, and a vacuum pumping station similar to the Pfieffer CombiLine
WU 1902 Roots pump with rotary vane backing pump (Se = 500 L/s), the pumpdown time for each tube section would be 1616 minutes or nearly 27 hours. There is
the possibility to decrease this time with custom designed and built vacuum stations
with higher pumping speeds. It is highly recommended for the airlock chambers to
design a custom pumping station with at least S0 = 8000 L/s in order to keep the
vacuumizing time less than 1 minute.
85
Chapter 8
Hyperloop Compressor
8.1
Introduction
A factor limiting the high speed movement of the Hyperloop capsule through a tube
containing air is the Kantrowitz limit. For a given tube to capsule ratio there is
a maximum speed that will choke the flow. If the capsule is too close to the tubes
walls then the capsule will eventually be forced to push the entire column of air in the
system. The Kantrowitz limit constrains the system to either go slowly or have a very
large tube diameter, neither of which are ideal. The proposed solution is to mount an
electric compressor fan on the nose of the pod that will actively transfer high pressure
air from the front to the rear of the capsule. Throughout the journey the weight of
the capsule will be supported by an air bearing system and this compressor will also
serve to supply the necessary air.
8.2
The processing of air through the alpha compressor design can be seen in Figure 8.1.
Using energy balance, mass and momentum conservation, and isentropic approximations this system can be analyzed. It was first discovered that by using the alpha
design, the incoming flow to the first axial compressor is too fast and turbulence
will cause the flow to become supersonic locally. It is very expensive to operate a
compressor when shock waves are forming inside so a diffuser is needed to slow the
flow down before the air is compressed. This was corroborated by a preliminary CFD
analysis the company ANSYS ran on the proposed Hyperloop capsule design.
86
8.3
Stagnation properties will be used for the analysis of a design with a diffuser in front of
the axial compressor. Assuming compressible flow through the diffuser and constant
specific heats the outlet pressure to the diffuser is
p02
1 2 1
= ptube 1 + d
M
2
(8.3.1)
where d is the diffuser efficiency (98%), M is the Mach number of the incoming
flow, and is the ratio of specific heats. Because the diffuser will be adiabatic, the
stagnation outlet temperature is
T02
1 2
= Ttube 1 +
M
2
87
(8.3.2)
Using the alpha tube conditions for the maximum capsule speed (ue, max ), p02 =
184.95 Pa and T02 = 350.6 K.
The mass flow rate of air through the compressor is limited by the capsule speed,
tube air density, and inlet area. The density of the air in the evacuated tube is
very low so there are less air particles available to the compressor. The maximum
reasonable inlet area is assumed to be 87% of the capsules frontal surface area. For
a tube pressure of 100 Pa and maximum capsule speed (ue, max ), the maximum mass
flow rate through the compressor (m c ) is about 0.49 kg/s. The pressure ratio of the
compressor can be found from the compression ratio (rp = r ). The alpha design
compression ratio, r = 20 will be used at first. The outlet pressure from the axial
compressor is then
p03 = rp p02
(8.3.3)
and the outlet temperature is
T03 = T02
1
1+
c
rp
1
(8.3.4)
where c is the isentropic efficiency of the compressor (85%). The power input is
given by
1
T02
Pmotor = m c cp
rp 1
(8.3.5)
c
Using the alpha parameters, it is found that p03 = 12.26 kPa, T03 = 1305 K, and
Pmotor = 470 kW. This configuration requires more power input and achieves a much
higher pressure than needed and a hotter temperature than desired. It is therefore
suggested that the compression ratio be lower than 20, especially since having the
diffuser raises the inlet pressure to the compressor. A compression ratio of 9 gives
a static compressor outlet pressure of p3 = 2.14 kPa and a static temperature T3 =
779 K, which are very close to the values listed in the alpha design. This lowers the
power input to 286 kW. Using mass conservation and assuming the area ratio of the
compressor outlet to inlet is 0.7, the exit velocity of the compressed air is 60.23 m/s.
After the first axial compressor, the air is cooled and about 60% of this air is
bypassed via a narrow tube near the bottom of the capsule to the capsules tail.
The nozzle expands the flow, generating thrust to make up for the losses due to the
aerodynamic drag. The remainder of the cooled air travels to a second compressor
where the air is compressed to the pressure desired for the air suspension system. This
air is then again cooled via an intercooler and then stored in an onboard pressure
88
8.4
The amount of water needed to cool the air at the two intercooolers was investigated.
Using the alpha document values for the water reservoir in Figure 8.1, the mass
flow rate of water needed to cool the compressed air to the desired levels using both
intercoolers was computed.
It is assumed the intercoolers achieve constant pressure cooling. Taking a look at
the first intercooler, the heat absorbed can be found from
Q absorbed = m
H2 O, (l) cp, H2 O (Tout, H2 O Tin, H2 O )
(8.4.1)
(8.4.2)
where Tcomp1, out = T3 , Tcomp2,in = 300 K, Tin, H2 O = 293 K, and Tout, H2 O will be determined by examining the second intercooler. The specific heats at constant pressure
for liquid water and air are cp, H2 O and cp, air , respectively. The mass flow rate of air
is the mass flow rate of air through the first compressor (0.49 kg/s). The mass flow
rate of the liquid water is unknown and will be solved for. Similar equations can be
established for the second intercooler using only the air mass flow rate though the
second intercooler (0.2 kg/s). The known parameters for the second intercooler are
the inlet and outlet temperatures of the air, which are, respectively, the outlet temperature from the second compressor (T5 = 510.5 K) and the desired temperature of
89
the pressure vessel air (Tvessel = 400 K). The output water temperature of the second
intercooler is assumed to be the temperature that will just make the phase transition
to steam (373 K). The inlet water temperature is equal to the outlet temperature of
the water from the first intercooler. This will then produce a system of two equations
with two unknowns (m
H2 O, (l) and Tout, H2 O ).
It was found that a much larger mass flow rate of water was needed than anticipated by the alpha document. The m
H2 O was computed to be 0.77 kg/s compared
to the alpha documents 0.14 kg/s (typical mass flow rate for intercoolers found in
racing cars). This is significant because the mass flow rate dictates the mass of water
each capsule will need to carry onboard. Assuming a total trip time of 35 minutes,
the mass of water needed to be stored would be 1617 kg (compared to 294 kg for the
alpha document). This mass increase (also container volume increase) is significant
and the effects will propagate throughout the rest of the design. For example, more
compressed air input will be needed for the air suspension system in order to maintain capsule levitation. This will mean the compressors will need to raise the pressure
slightly higher. This will correspond to more energy input as well as a higher output
temperature from the compressors. The higher output temperatures will then need
even more water to cool to the desired temperatures, increasing the onboard water
mass further. Iterations on these calculations will need to be run when implementing
a full design.
Upon closer inspection, the mass flow of liquid water presented in the alpha document can be achieved if the temperature of the steam is allowed to be higher. A
water mass flow rate of 0.14 kg/s will be required if the temperature of the steam
is 730 K. Further investigation is needed to determine if this higher temperature of
steam would be a problem. In addition, the possibility of not cooling the bypassed
air to the capsule tail should be explored. If this air is not cooled, then less air needs
to be cooled and less mass of water coolant is required.
The steam reservoir needs some more attention than given in the alpha document.
The volume requirements of the steam reservoir will be much larger than those of the
water reservoir because steam has a much lower density. The enthalpy of steam is
also much higher, and if the steam is needed to be compressed before storage this will
require additional energy input. The alpha document suggested the steam reservoir
will be swapped out after the journey, but provided no volume specifications. There
may be a better method for a heat sink than using water. Certain metals will be
more effective, but at the cost of larger mass. These alternative heat sinks should be
explored.
90
8.5
Summary
The primary purpose of the compressor is to prevent the flow from choking. In
addition, the compressor will supply high pressure air to the air suspension system as
well as provide some additional thrust at the rear of the capsule. A major adjustment
to the alpha compressor design was found to be adding a diffuser in front of the
first compressor. This will help prevent the air from going supersonic locally at the
compressor. In addition, this would also compress the air before the inlet to the
compressor, lowering the required compression ratio and motor input power required.
Many assumptions made by the alpha design were used in the evaluation of this
design. Many of these parameters could be optimized based on the analysis of the
other subsystems. For example, the amount of air bypassed to the tail of the capsule
and the amount routed to the air suspension system should be investigated. The
amount of air required by the air suspension system will be the driving factor. It
was found that at the low pressure of the tube that the aerodynamic drag is not
large enough to significantly slow the capsule so the amount of thrust gained back
from the nozzle should only be considered as a bonus, not a major design parameter.
Preventing the airflow from going supersonic will put a limit on this thrust generated.
It is recommended that this be investigated in future designs. The heat released by
the capsule compressing air, then expelling it downward and backward has raised
concern from many critics. It appears the water cooling system will need to have a
larger flow rate than originally anticipated, but further research is needed.
91
Chapter 9
Modeling the Hyperloop Route
9.1
Introduction
The proposed alpha route from Los Angeles to San Francisco will maintain a path as
close as possible to the established right-of-way while respecting existing structures to
reduce capital costs. This route will follow I-5 and I-580 for the majority of the 563 km
journey with the tube construction in the median. There exists the possibility to have
more stations and split tracks along the way, but these will not be considered at this
time. The capsule will carry 28 passengers and depart on average every two minutes
according to the alpha document. To ensure passenger comfort, lateral acceleration
is limited to 0.5g and longitudinal acceleration is limited to 1g during propulsion.
Propulsion stations will be optimized to ensure appropriate g-force is experienced by
passengers and to maintain speed. The absolute max speed is limited by the condition
by which the flow will become sonic and form shock waves (assumed to M = 0.99).
The other constraint on the capsules speed is imparted by the bend radius of the
route so passengers do not experience lateral accelerations greater than 0.5g. In some
places the route will need to be altered from the highway in order to mitigate g-force
concerns. Different sized pylons and tunneling will also be used to minimize elevation
changes.
This chapter looks into the Hyperloop route focusing on the capsules speed and
position over the trip length, as well as overall trip time. The route path is modeled
for the propulsion stations that will keep g-forces at a comfortable level and using the
aerodynamic drag force derived in Chapter 3 to get the systems dynamics.
92
9.2
Route Visualization
9.2.1
Capsule Deceleration
The capsules deceleration was examined in the absence of the reboost propulsion
stations for the three cruising speeds using the aerodynamic force model from Chapter
3 and solving the differential equation established by Newtons second law
x =
Ftot
mc
(9.2.1)
where Ftot is the total aerodynamic drag force, and mc is the capsules total mass. This
differential equation is solved to find the capsules position and velocity as functions
of time (Figure 9.1 and Figure 9.2 respectively) due to the aerodynamic drag forces
(Figure 9.3) acting on the body. The power loss (kW) due to the aerodynamic drag
forces is plotted in Figure 9.4 and ranges from 18 to 260 kW.
Hyperloop Deceleration Capsule Position
800
700
600
Position (km)
500
400
300
200
100
0
0
500
1000
1500
2000
2500
Time (s)
93
300
Speed (m/s)
250
200
150
100
and gap
and gap
and gap
mid
low
low
min
max
min
50
0
500
1000
1500
2000
2500
Time (s)
9.2.2
Throughout the Hyperloops journey the path will need to turn. To ensure passenger
comfort, turns will need to be taken at a maximum of 0.5g of lateral acceleration.
There is the possibility of banking the capsule in the tube during turns, but it is
assumed this type of system is not in place for this analysis. Table 9.1 contains the
minimum bend radius for turns at each of the three cruising speeds (ue, max , ue, mid ,
and ue, low ) using Equation (9.2.2). These results confirm the suggestions found in
the alpha document.
rmin =
94
u2e
0.5g
(9.2.2)
100
200
300
400
500
600
and gap
max
min
700
and gap
and gap
and gap
mid
low
low
min
max
min
800
0
500
1000
1500
2000
2500
Time (s)
and gap
and gap
max
250
max
max
min
mid
and gap
min
200
low
and gap
min
150
100
50
0
0
500
1000
1500
2000
2500
Time (s)
Figure 9.4: Power loss due to aerodynamic drag force acting on Hyperloop capsule
as it decelerates.
95
9.2.3
ulow = 134.11
umid = 248.11
umax = 339.75
3.67
12.6
23.5
The route considered features three propulsion stations to accelerate the vehicle after
leaving Los Angeles and three propulsion/regenerative braking stations to decelerate the vehicle before arriving in San Francisco. Undoubtedly, top speed should be
reached as soon as possible and maintained for as long as possible to minimize total
trip time. However, the track path puts limitations on the vehicle speed for certain
sections of the track. If the route is to follow the interstate system and respect existing structures, the bend radius of the track will be much smaller at the start and
end of the journey as the roadway makes many turns through the suburbs. The defined bend radius of the roadway and the condition of 0.5g lateral acceleration puts
a constraint on the capsule speed. The middle section features right-of-way which is
much straighter, and thus faster speeds can be reached. The alpha document lists
the distances along the route that need to be traveled at the low speed, at the mid
speed, and at the high speed. In addition, it lists the times after departing L.A. when
the capsule reaches a propulsion/braking station (Table 9.2).
Table 9.2: Alpha route segment times and distances traveled since leaving L.A.
Track Segment
ulow Depart.
umid Depart.
umax Depart.
umax Arriv.
umid Arriv.
ulow Arriv.
0
21.60
86.00
480.75
503.06
563.00
96
First, the suggested alpha section times for the route are used to evaluate the
aerodynamic drag force model used by the alpha document staff compared to the
model developed in Chapter 3. A plot of the capsules position as a function of time
since departure from L.A. will reveal if the capsule will actually arrive in San Francisco
using these suggested propulsion times. Then, using the alpha section distances
corresponding to each speed the Chapter 3 aerodynamic analysis will be used to carry
out the journey to completion with an arrival in San Francisco. Figures 9.5 and 9.6
show the speed and position plots as functions of time since departure in L.A. From
this analysis it was found that using the alpha document timing recommendations the
capsule would not actually reach San Francisco. The final position for this analysis
is only 534.8 km compared to the total trip distance of 563 km stated in the alpha
document. Optimizing the journey for the given speed sections lengths, the journey
is visualized to completion. The differences between the blue (alpha times) and red
(required alpha section lengths) reveal that the alpha times do not allow the capsule
to coast at maximum speed for long enough, and thus the capsule would not reach San
Francisco under these conditions. The total time to get from L.A. to San Francisco
using the alpha segment distances and the aerodynamic drag model from Chapter 3
is 2181 seconds (36.35 min). This is 47 seconds longer than the total time proposed
in the alpha document (2134 seconds). Compared to the Califronia High Speed Rail
trip time of 2 hours 38 minutes, the time saved is huge.
97
300
Velocity (m/s)
250
200
150
times
distances
100
50
0
0
500
1000
1500
Time from L.A. (s)
2000
2500
Figure 9.5: Speed of Hyperloop capsule as a function of time from L.A. departure.
500
Position (km)
400
300
times
distances
200
100
0
0
500
1000
1500
Time from L.A. (s)
2000
2500
Figure 9.6: Position of Hyperloop capsule as a function of time from L.A. departure.
98
9.2.4
During peak business hours the Hyperloop will want to maximize the number of
passengers transported. The minimum departure delay between two capsules will
ensure there is never a collision and is calculated under the assumption that the the
trailing capsule can stop as soon as there is an emergency with the leading capsule.
The minimum departure delay for 1g deceleration is about 27 seconds. The minimum
acceleration needed to stop under a range of departure delays is compiled in Table
9.3. Figure 9.7 plots the distance between two capsules as a function of time since
departing L.A. for these same departure delay times. The minimum deceleration
curves are overlaid on these plots (dotted red lines) to show what segments in the
journey impose these minimum deceleration for each departure delay time. The
regions where each blue curve and its corresponding minimum deceleration red curve
almost intersect is circled in green. It can be seen this only happens in the first two
segments of the journey. Once the first capsule gets to full speed, it is much more
difficult for the second capsule to ever catch up.
Minimum Deceleration
27
60
90
180
300
1g
0.420g
0.275g
0.135g
0.085g
99
100
20
40
60
80
100
120
500
Limiting region
180s
300s
90s
60s
1000
1500
Time Since Second Capsule Departed From L.A. (s)
Coresponding min
deceleration
27s
2000
2500
Figure 9.7: Distance between two capsules as a function of time since the second capsule departed L.A. plotted for multiple
departure delay times. Minimum deceleration is overlaid (dotted red curve) to indicate regions where close contact would occur
(green circled areas).
101
5000
4000
20 Pax/Capsule
28 Pax/Capsule
40 Pax/Capsule
3000
2000
1000
0
30
60
90
120
150
180
210
Departure Delay Time (s)
240
270
300
Figure 9.8: Number of passengers per hour as a function of departure delay with
capsule capacities 20, 28 (), and 40 passengers.
Number of Hyperloop Capsules vs. Departure Delay
250
200
150
1 min Load Time
5 min Load Time
10 min Load Time
100
50
0
50
100
150
200
Departure Delay Time (s)
250
300
Figure 9.9: Number of capsules required for a given departure delay time with load
times 1 min, 5 min, and 10 min.
102
9.3
The track bend radius defined by the existing roadway limits the capsules speed in
certain sections of the track in order to maintain passenger comfort. Earlier, the bend
radii of each section and the corresponding speed was taken from the alpha document
recommendations. Now, the track path is looked at more closely to observe the lateral
g-force on passengers throughout the trip. Using geographical data points of the route
taken from Google Earth and computing the bend radius from parabolic curve fitting
estimates, the lateral acceleration over the trip could be plotted.
The bend radius of a curve is defined as
3/2
(t) =
|1 + f 0 2 (t)|
|f 00 (t)|
(9.3.1)
where f (t) is the route path. The equation for the route path can be estimated by a
parabolic curve fitting method for a group of route points. The lateral acceleration
can then be found from
u2e (t)
(9.3.2)
ag,trans =
(t)
Using a technique developed by MathWorks to eliminate redundant route data
points and lightly smooth the curve, the following lateral g-force plot can be developed
(Figure 9.10) [4]. It can be seen there are several sections that exceed 0.5g where the
Hyperloop alpha document listed this would not happen. The route path would
need to be altered in these sections, requiring more land to be bought and potential
destruction of existing structures. In addition, it is apparent the lateral g-forces can
change rather rapidly in the suburban sections. Although most of the accelerations
are in the comfortable range, humans still do no like the constant acceleration changes.
This transportation system will not be very successful if it mimics a roller coaster.
A section of track that requires smoothing was examined in order to evaluate how
much land may need to be acquired to make a more comfortable ride. While moving
away from the interstate in the middle section of the journey will not be difficult, this
could be very challenging in the suburban sections without encroaching on private
property and displacing current infrastructure. High speeds may need to be sacrificed
in the suburb regions. Total trip time will decrease, but passenger comfort will be
maintained. An alternative is to have more active braking and propulsion during
the suburbs, but this would add extra costs as well as potentially more passenger
discomfort due to constant accelerations and decelerations. Maglev allows for easier
103
0.5
0.5
1
0
10
15
20
Time (min)
25
30
35
40
Figure 9.11: Example bend radii for route section with suggested low speed (ue, low )
and corresponding minimum bend radius (rmin ) of 3.67 km.
(b) Area of land needed to acquire for making track perfectly straight.
Figure 9.12: Track smoothing and required land area. I-5 is displayed in blue, while
the smoothed track is displayed in yellow.
105
Figure 9.13: San Francisco surburban area where route path needs to be altered. The
minumum amount of curvature needed for ue, low is a rmin of 3.67 km (displayed in
red). Alternatively, this suburban section can be diverted completely by moving the
track north (yellow curve).
106
Figure 9.14: San Francisco surburan area that shows the minimum necessary altered
path (yellow) to travel at ue, low . This path goes straight through The Neigborhood
Church.
Estimating Costs
Estimating the capital costs associated with acquiring the land needed to make the
Hyperloop route smoother and more comfortable for passengers while reducing overall travel time is very difficult. Land costs are expected to be cheaper in the middle
section of the route where there is less established civilization. Obtaining the land
here may be as simple as purchasing from an owner and then receiving a permit. The
suburban areas are where it gets difficult to quantify the land acquisitions. The land
will be much more expensive and the politics become much more complex. Many
businesses and communities may not want to be displaced for the Hyperloop regardless of the amount of money offered to them. On one hand the California High Speed
Rail project has estimated $3.915 billion in capital costs to purchase real estate and
a total of $13.059 billion for sitework, right-of-way, land, and existing improvements
[8]. On the other hand, the Hyperloop alpha document has only estimated $1 billion
for permits and land. The reasoning for this low estimate is that since the Hyperloop
will be elevated on pylons it can be placed in the median of the existing interstate system while only deviating from this roadway when necessary to reduce lateral g-forces.
However, the alpha document completely ignores all the politics behind working with
107
9.4
Summary
It was found that the aerodynamic analysis of the Hyperloop used by the alpha document team was too conservative. Using the aerodynamic model derived in Chapter
3, the revised total trip time was found to be 36.35 minutes (47 seconds longer than
the alpha time). To be able make emergency stops at 1g without collisions, capsules
can depart roughly every 30 seconds. For capsules with a capacity of 28 people, this
corresponds to 3,722 passengers per hour each way and a requirement of 155 191
capsules. Taking a closer look at the lateral accelerations throughout the route, it
was determined significant alterations from the interstate system will need to be made
in order to maintain passenger comfort (lateral accelerations below 0.5g and less frequent changes in acceleration). Large areas of land will need to be acquired to smooth
the route path. This process will encounter more obstacles in suburban areas due to
more existing infrastructure and resistance to displacement. The political and community effects surrounding land acquisition may be enough to delay the project and
significantly increase the total costs. Solutions to managing the route in suburban
areas include route detours, more frequent braking, lower speeds, and tunneling. The
degree to which these route problems can be solved will be a driving factor of total
costs and project scheduling.
108
Chapter 10
Hyperloop Infrastructure
10.1
There are four possible tube route layout schemes for the Hyperloop. The tube can
be built overhead on pylons, at ground level, shallowly below ground, or deeply below
ground. Each method has its advantages and disadvantages. A study from Xijing
University in China investigated the trade-offs between the different schemes [65].
The overhead scheme was the clear winner. While the underground methods are
advantageous in respect to land acquisition and providing robustness to temperature
and climate fluctuations, they are not realistic due to extremely high construction and
maintenance costs. Tubes of the same size diameter would be much more expensive
underground. Especially, since the costs of tunneling are much higher than those of
building bridges. In addition, it is very difficult to check the vacuum state of the
tube and then to fix a leak. The Swiss company SwissMetro proposed a completely
underground vactrain that would be 50 meters below street level to connect Bern to
Zurich in 12 minutes [44]. Even if these issues could be minimized and costs reduced,
it would be hard to convince people to ride in a small tube embedded in the rock
deeply underground where there is no easy escape in the case of an emergency. It is
not surprising this concept has not gained much traction. Thus, the above ground
solutions are more favorable.
The ground and overhead options will share many of the same advantages and
disadvantages. The overhead option will face slightly higher construction costs due to
pillar construction as well as some degree more of maintenance challenges. However,
the estimated capital saved acquiring less land than the ground method gives the
109
overhead method the overall edge. Furthermore, tubes on the ground would be more
exposed to damage by humans and natural occurrences like floods. Studies conducted
for current Maglev projects also have corroborated that overhead transportation is
optimal. Further civil engineering evaluation should be conducted for the specific
region of interest. Factors such as the seismic activity in the California area could
swing the scheme in one direction or the other.
10.2
There are two viable material/structure options that have been proposed with low cost
and high reliability for evacuated tube transportation. The first is a pure steel tube.
The second is a concrete composite structure with a thin metal layer. Differences
concerning the manufacturing and construction of tubes with both materials will be
discussed.
Technologies that depend on maintaining a vacuum environment typically use
steel because it has good strength characteristics as well as low permeability to keep
the tube sealed. The current price of steel is around $800 - $1000 per ton [11].
Concrete has low cost ($70 per ton), high compressive strength, and good stiffness,
but it will not maintain the vacuum environment to the same degree as steel because
it is porous [63]. However, gas-tight concrete has been used in nuclear reactor pressure vessels and seawater desalt jars for higher expenditures. The type of composite
structure needed would feature a thin inner steel wall with an outer concrete layer.
Because the inside steel layer will want to be pushed away from the concrete layer,
special attention needs to be directed to securing it in place. This may be a challenge to avoid. The overall thickness of the tube may be larger with the composite
structure, but the cost will be less since concrete is much cheaper than steel. This
structure will have large compressive strength to allow for a more intense vacuum
degree. A stainless steel layer could also be added to the outside of the concrete to
help mitigate damage and corrosion. This layer would tend to be pulled against the
concrete layer, making the fastening much easier.
10.3
distribution throughout the thickness will not vary significantly so it can be assumed
to be uniform [24]. Cylindrical vessels are subject to Hoop stress (circumferential
stress) from the pressure differential established by the vacuum. The Hoop stress can
be calculated from the following formula:
Hoop =
pg rtube
t
(10.3.1)
where pg is the gauge pressure, rtube , is the inner radius of the tube, and t is the tube
thickness.
To design the tube thickness, the hoop stress multiplied by a factor of safety (FoS)
is set equal to the compressive strength (B ).
t=
FoS pg rtube
B
(10.3.2)
For reasonable steel choices, B ranges from 180 MPa (stainless) to 1200 MPa
(low carbon martensitic steel)[24] [11]. Using a FoS = 2, a tube with a diameter
equal to 2.23 m (alpha diameter), the tube thickness needs to only equal 1.3 mm to
0.2 mm, depending on the steel chosen. This, however, is much smaller than what
actually needs to be designed. Thin walls are subject to torsion deformation and
destabilization. Zhang and Oster suggested tube thicknesses of 8-15 mm for a tube
with diameter 2.4 m [63].
The tube should be designed to handle longitudinal stresses. This includes the
bending stress induced by thermal expansion, bending stress induced by weight of
tube (will be minimal since tube is simply supported), and longitudinal stress induced
by internal pressure [17]. The bending stress is equal to
BL =
32DM
M
=
I
((D + t)4 D4 )
(10.3.3)
6EIL
L2
(10.3.4)
where E is the Youngs Modulus, I is the area moment of inertia, L is the length of
tube considered, and
L = l LT
(10.3.5)
111
The average temperature change in a day for the Los Angeles area in 2013 was 8.3 K.
Assuming the coefficient of thermal expansion for steel is l = 13 x 106 m/m-K, the
maximum L can be computed. For a 30 m long section of tube (distance between
two pillars), the maximum change in length is only 3.237 mm along its length. For a
20 mm thick tube the expansion radially is only 0.003 mm. Thermal expansion will
not be a major concern for the safety of the tube wall as long as expansion joints are
incorporated throughout the length of the tube.
The shear (torsional) stress the tube must withstand is
T =
TD
32 T D
=
2J
((D + t)4 D4 )
(10.3.6)
Expected torsional moments, T , can be plugged into this equation and used to check
the thickness of the tube selected. Due to limitations in analysis time, a complete
structural analysis with simulations was not completed. It will therefore be assumed
the tube thickness is 23 mm for a tube of diameter 2.23 m (alpha conditions) and will
scale with a change in tube diameter in accordance to the Hoop stress relationship.
For example, a 6 m tube would correspond to a 65.53 mm steel tube thickness.
10.4
The alpha document presented a basic structural model for the pylons needed to
support the Hyperloop tube and mounted solar arrays. Reinforced concrete pylons
will support and constrain the Hyperloop tube in the vertical direction, while allowing
longitudinal slip for thermal expansion as well as dampened lateral slip to mitigate
the effects of earthquakes. Less frequent expansion joints are needed than ground
structures because the tube does not need to be rigidly fixed at any point. Two
adjustable lateral (XY) dampers and one vertical (Z) damper can be placed inside
the pylons, isolating the ground movements and thus, mitigating earthquake risk.
The thermal expansion slips will be placed at stations where speeds are lower. The
distance between pillars will be 30 m to reduce tube deflection effects and increase
resistance to seismic loading. There will be roughly 25,000 pillars with a nominal
height of 6 m. The alpha document team performed limited structural simulations
on the model described and claimed the capability to withstand atmospheric pressure,
tube weight, and earthquakes. These simulations were minimal and more in-depth
modeling/simulating needs to be done. Specifically, more attention needs to be spent
112
on the effect of earthquakes. It is likely more analysis will reveal the need for more
dampers and expansion joints in more frequent concentration along the tube length.
Earthquakes can be designed for, but this design will likely contribute to higher
costs. The Trans-Alaska Pipeline runs across the Denali Fault and was designed to
handle a maximum expected event reported from geological studies. When a 7.9
magnitude earthquake struck the pipeline in 2002, the design held and not a single
drop of oil spilled. While a few supports were damaged, the piping remained intact.
The engineering solution to accommodate the projected fault movement and intense
earthquake shaking was to support the piping on Teflon shoes that are free to slide on
long horizontal steel beams. This only added $3 million to the total cost when built
in the 1970s, and saved potentially $100 million in repairs [60]. While air molecules
may more easily escape a damaged evacuated tube than oil molecules, earthquakes
risk can be designed for in the Hyperloop. Furthermore, earthquake risk applies
to everything from transportation to buildings. With any type of infrastructure in
earthquake zones, some risk has to be accepted and expected damage accounted for
in its lifetime.
10.5
Constructing the Hyperloop will be a massive engineering project. As with any large
project there will be unaccountable factors and unknowns that appear throughout
the projects development. The Hyperloop is likely to encounter similar setbacks to
what caused Californias $68 billion high-speed rail system to go far over budget.
Political and community pressures will be major players. Lessons can be learned by
examining past projects that have both succeeded an failed.
10.5.1
The Princeton Plasma Physics Laboratory (PPPL) shared some engineering cost lesson learned from the challenges of their National Compact Stellarator Experiment
(NSCX) [55]. Technical challenges manifested themselves throughout all phases of
the project (design, R&D, fabrication, assembly) and led to the Department of En-
113
ergy (DOE) ceasing to fund the project. The factors contributing to the cost increases
included complex geometry, tight tolerances, material requirements, and performance
requirements. The magnitude of cost growth from the project baseline was investigated. It was found that there was 3.0 factor increase for design, a 1.3 increase for
R&D and prototyping, a 2.3 increase for fabrication and assembly, and a 1.8 increase
for procurement. The total factor of increase was 2.0. Geometry accounted for 35% of
the increase, while accuracy accounted for 37%. Very tight assembly tolerances led to
the need for sophisticated metrology equipment and metrology dedicated technicians,
engineers, and analysts. The tolerance requirements were not well understood when
the budget was established because certain components were not prototyped.
The lessons learned from this project will apply to the Hyperloop. A premature
establishment of a firm cost and schedule baseline should be avoided until a more
mature design and fabrication base can be developed. Tight tolerances may be necessary with Hyperloop tube due to the very small suspension gap height. These
tolerances should be applied surgically and only where they are critical. Tolerances
drive vendor procurement costs, require in-house engineering time to disposition nonconformance reports (NCRs), and increase assembly time. Labor costs are not cheap
and large management may be required in addition. The analysis time to determine
if the measured variation is meaningful is often underestimated. Prototyping is critical to account for all aspects of assembly. Simplifying the design can mitigate cost
growth. Finally, it is important to recognize the nature of high tech/high risk projects
and avoid prematurely establishing cost and schedule baselines until a more mature
design/fabrication experience base is established. The government and most other investors are very risk adverse so when costs of a new technology start to exponentially
increase people will begin to lose faith and may end up canceling the project.
While this list is not exhaustive, there are other factors related to the Hyperloops
infrastructure. The tight tolerances are subject to temperature change so analysis
will need to determine if variation is problematic. Compliance needs to be built into
the design to ensure a smooth assembly process. Corrosion, espcially around joints,
can also cause harm. The Keystone Oil-Pipeline accounts for 0.1 mm/yr in corrosion
(80-85% in the joints)[17]. While running oil is significantly different than a capsule moving through air, corrosion still needs to be investigated. Since California is
an earthquake zone, everything moves and much more detailed studies need to be
conducted than the simple structural analysis presented in the alpha document. Environmental studies need to be conducted to ensure the areas ecosystem is preserved.
Community members may also be upset by the visual impact created by a Hyperloop
114
system. The next professional design evaluation of the Hyperloop needs to factor in
all the risks and unknowns. This was not considered in the alpha proposal.
10.5.2
Large-scale infrastructure projects bring substantial impacts on communities, environment, and budget. While costs are astronomical, they can help boost local economics and create jobs for thousands of people. Environmental issues will contribute
to the political sphere. Large infrastructure projects almost always cost more than
the initial projections. A 2002 Danish study by Bent Flyvbjerg, Mette Holm and
Soren Buhl looked at 250 major infrastructure projects dating back to the 1920s and
found that roughly 90 percent of them went over budget by an average of about 45
percent [18].
Oil Pipelines are examples of large-scale infrastructure projects with similar structures to to the Hyperloop tube, but with less performance constraints. The TransAlaska Pipeline consists of a 1.22 m diameter piping system spanning 1,287 km with
11 pump stations. In 1969, the proposed cost was listed to be $900 million [32]. The
final construction cost totaled $8 billion, while enlisting the work of 70,000 people.
The entire Keystone Pipeline project (the existing 2,615 km plus the 1,897 km of
KXL) is projected to cost $13.2 billion to build [12]. TransCanada has stated that
$6.2 billion has already been used to establish the current infrastructure and $7 billion more is needed to complete KXL. The completion of this addition is currently
a controversial issue. These oil pipelines mainly build on open land where existing
structures will not interfere. The Hyperloop route will not have as much luxury,
especially when approaching urban areas.
To examine the costs of massive infrastructure projects in heavily populated areas,
the Big Dig (Central Artery/Tunnel Project) in Boston, Massachusetts was examined.
The Big Dig rerouted the Central Artery (I-93) through the heart of the city by using
a 5.6 km tunnel and establishing a wider roadway. In addition, the project included
the construction of the worlds widest cable-stayed bridge, the deepest underwater
connection in North American, extensions of I-90, and many other infrastructure
refinements. The project commenced in 1991 and it proved to be on of the most
technically challenging urban transportation developments in the U.S. Originally estimated to cost $2.8 billion ($6.0 billion with inflation) and be finished by 1998, upon
completion in 2007 the project was estimated to have cost $14.8 billion [35] [46]. The
actual cost was released to the public in 2012 with a total more than $24 billion,
115
earning the title as the most expensive highway project in the U.S. The scale of this
project compares to the Panama Canal and the Channel Tunnel. While this example
may be on the extreme side, it does show how dealing with a massive project in
heavily populated areas can lead to significant cost and time increases. There will
always be unknown variables that are not accounted for at the projects start, and
dealing with the problems that arise will be more challenging in urban sections. The
Hyperloop will not have much trouble in between Los Angeles and San Francisco, but
as the route enters either city there is a whole array of different factors to consider
and the challenges they bring will be hard to predict.
10.6
10.6.1
Tube Costs
It is assumed the tube will be constructed of steel costing $800 per ton. The alpha
document tube has a diameter of 2.23 m and a thickness of 23 mm. To examine the
effect of increasing diameter, the tube thickness will scale with diameter defined by
the Hoop Stress relationship. Estimates of welding and installation cost is taken to
$800,000 per km [63]. The following raw material costs plus welding and installation
costs for the Hyperloop tube can be found in Figure 10.1 as a function of tube diameter. These estimates for raw material costs plus welding and installation costs
are larger than the extrapolated alpha costs for tube construction. For the alpha
documents passenger system diameter, the revised cost is $1.1 billion compared to
$650 million for the alpha. These new cost estimates may still be conservative. Tooling can rapidly increase with scale factor, so the tube costs may increase with D3
instead of only D2 . In addition, the cleaning and boring machine discussed by the
alpha document to smooth the inside of the tube will need to be custom made, and
depending on the tolerances desired this may add a substantial expense.
10.6.2
The alpha document predicts pylon construction will cost $2.55 billion. This likely is
an underestimate because it is assuming these pylons are identical pieces that can be
mass produced and dropped in place. Since the landscape varies rather significantly
in this region the pylons will need to vary more in order to keep the vertical accelerations experienced by the passengers constant, requiring more custom manufacturing.
116
3
2.5
2
1.5
1
Steel Material, Welding & Installation
0.5
Alpha Projected
0
1.5
2.5
3
3.5
Tube Diameter (m)
4.5
Figure 10.1: Tube costs as a function of tube diameter. The sum of the pure steel
material costs plus welding and installation costs is plotted in blue. This is larger
than the alpha projected tube construction costs (red).
The California High Speed Rail anticipates nearly ten times the cost for viaducts
(pylon structure) than the alpha document [8]. The CHSR lists an average of $44.39
million/km (including contingency) for viaducts versus $4.53 million/km for the Hyperloop. It should also be considered that after a budget review from 2009 to 2012
the CHSR increased the budget for viaducts by more than a factor of 2 [9]. However,
it is hard to compare the Hyerloop pylons to the CHSR because their structures have
fundamental differences. The high speed rails viaduct will need to support significantly more mass. The Hyperloop will feature similar superstructure (below ground)
to the CHSR, but the substructure (above ground) will be different. It is a good
assumption to consider the Hyperloop pylons will cost somewhere between the alpha
proposal and the CHSR 2012 viaduct budget, with the revised cost being closer to
the alpha document end. It is then assumed the pylon cost will be 3 times the alpha
projection due to more customization than originally anticipated and the evidence
that original projected costs will at least increase by a factor of 2 from initial budget.
Therefore, a suggested estimate for pylon costs is $13.59 million per km for a total of
$7.65 billion for the entire Hyperloop pylon construction. This is already larger than
117
the total alpha cost of $6 billion. It is also assumed that pylon cost is only a weak
function of tube mass.
10.6.3
Tunneling Costs
Tunneling for the Hyperloop route can be expensive. The alpha document estimated
24.2 km of tunneling for where extreme local gradients are greater than 6%. The
estimations state tunnel construction will cost $600 million ($24.8 million per km).
The California High Speed Rail estimates tunneling to cost a total of $13 billion
($168.32 million per km), more than double the total from the 2009 business plan
(though the cost increase per km was only about 6%)[9]. The Hyperloop may find it
will need to tunnel underground in the urban areas to avoid displacing infrastructure.
This will increase costs similar to what was seen from the CHSR 2009 budget to
the 2012 budget. Still, the Hyperloop tunneling costs will be much less than those
of the CHRS due to the much smaller bore size needed. As a result, less concrete
reinforcements will be needed. The CHSR investigated the methods of tunneling.
They found there is not a significant cost difference between using tunneling bore
machines and drill/blast methods. However, the drill/methods take longer so a tunneling bore machine would be preferred. Ted Zoli, National Bridge Chief Engineer
at HNTB Corporation, discussed how the tunneling for the Hyperloop could be done
with a custom horizontal directional drilling (HDD) unit [1]. While the largest used
can only bore out 64% of the required Hyperloop diameter, a custom unit could be
built. Still, the question remains if the time and costs to make a larger device will be
less than the alternate drill/blast method. Doubling the length of tunneling needed
for the Hyperloop and adding the cost of developing a larger boring machines, it is
suggested the total tunneling costs will be around $1.5 billion ($31 million per km).
This cost will scale with the area needed to bore for the given tube diameter.
There are several other unknowns that can contribute to the civil infrastructure
costs. For example, the alpha document suggests the Hyperloop will not cross the
San Francisco Bay. If this will be added to the route, there will need to be either
a bridge or a tunnel underneath the waterway. For now, this expenditure will be
ignored to keep cost evaluation consistent with the alpha document.
118
10.7
The total civil infrastructure costs were broken down and analyzed using other largescale infrastructure projects. These costs consist of the tube construction costs, the
pylon construction costs, and the tunneling costs and are plotted in Figure 10.2. For
the alpha tube diameter of 2.23 m, the total civil infrastructure costs alone reach a
total of $10.2 billion, 2.7 times the same cost predicted by the alpha document. Even
without considering the other capital costs, this is larger than the alpha documents
predicted $6 billion total capital expenditure for the entire Hyperloop system. The
other capital costs that need to be consider include everything associated with the
capsule, propulsion, solar panels and batteries, vacuum pumps, and land acquisition.
This will be discussed in the next chapter.
16
Tunnel Construction
Pylon Construction
14
12
10
8
6
4
2
0
1.5
2.5
3
3.5
Tube Diameter (m)
4.5
Figure 10.2: Total civil infrastructure capital costs as a function of tube diameter.
119
Chapter 11
Effect of Changing System
Parameters on Overall Trip Time
and System Costs
11.1
Introduction
Using the analyses from the previous chapters, the effects of changing several system
parameters on overall trip time, system operational costs, and system capital costs
were examined. Alpha document conditions were held constant while the selected
parameter was allowed to vary. The assumptions for the route used to calculate the
total trip time are the same as in Chapter 9.
11.2
The tube diameter proposed by the alpha document is very small (only 2.23 m). Many
humans may find it uncomfortable to ride in a capsule that is only 1.35 m tall. The
alpha document also discussed the possibility of building a larger system that could
transport cars as well as passengers. First, the effect of increasing the tubes diameter
on the overall trip time is investigated holding the capsule size and everything else
constant. Second, the possibility of creating a larger Hyperloop system which may
be more comfortable for riders and even be able to transport cars is explored. In the
latter case the capsule will scale with the tube diameter (constant blockage ratio).
120
Overall trip time will not be impacted as much as capital and operational costs.
38
37.5
37
36.5
36
Max Gap
Min Gap
35.5
35
0
0.1
0.2
0.3
0.4
Blockage Ratio
0.5
0.6
0.7
of the capsule change (maintaining constant aspect ratio), while the capsule length
remains constant. From Figure 11.2 the total trip time actually decreases slightly as
the overall system size is increased. This suggests that the the increase in mass of
the capsule outweighs the increase in total aerodynamic drag force, and thus there
are slower changes in accelerations.
Total Trip Time as a Function of Tube Diameter
39
38.5
Max Gap
Min Gap
38
37.5
37
36.5
36
35.5
1.5
2.5
3
3.5
Tube Diameter (m)
4.5
122
3000
2500
2000
1500
1000
500
0
1.5
2.5
3
3.5
Tube Diameter (m)
4.5
Figure 11.3: Propulsion energy as a function of tube diameter. The capsule size and
mass scale with diameter. The four propulsion segments are plotted.
11.3
Longer capsules may be desired to transport more passengers, impacting both trip
time and costs.
37.4
37.2
37
36.8
Max Gap
Min Gap
36.6
36.4
36.2
36
20
30
40
50
60
70
Capsule Length (m)
80
90
100
124
2000
300760mph
300550mph
550760mph
0300mph
1500
1000
500
0
20
30
40
50
60
70
Capsule Length (m)
80
90
100
Figure 11.5: Propulsion work as a function of capsule length for max suspension gap
height. The four propulsion segments are plotted.
with larger capsules is the increased mass will require a larger lift force for suspension. This will require more compressed air input from the air bearing suspension
system. There will be a limit to the amount of air supplied as well as increased power
requirements for the compressor machinery.
125
160
140
120
100
80
60
40
20
20
30
40
50
60
70
Capsule Length (m)
80
90
100
11.4
The tube pressure is one of the most important parameters in the Hyperloop system. Evacuating the tube to a low enough vacuum degree will significantly decrease
aerodynamic drag and allow the capsule to coast and maintain speed for most of the
journey.
140
120
100
80
60
Max Gap
Min Gap
40
20
0
2000
4000
6000
Tube Pressure (Pa)
8000
10000
38.5
38
37.5
37
36.5
Max Gap
Min Gap
36
0
100
200
300
Tube Pressure (Pa)
400
500
Figure 11.8: Total trip time as a function of tube pressure focusing on pressures 0 to
500 Pa.
127
700
300760mph
300550mph
550760mph
0300mph
600
500
400
300
200
100
0
2000
4000
6000
Tube Pressure (Pa)
8000
10000
128
11.5
The effect of varying the suspension gap height from 0 - 5 mm was analyzed under
the assumption there is always enough compressed air to keep the capsule levitated
in this gap range.
50
45
40
35
0
2
3
Gap Height (mm)
Figure 11.10: Total trip time as a function of gap height. Knee bend in the curve
occurs around 0.5 mm.
129
11.6
Other effects
Changing the gap temperature (gap viscosity) led to a very small increase in trip time.
The importance of this parameter has more effect on the materials of the suspension
blades.
11.7
Total Costs
The total capital costs presented by the alpha document have been revised throughout this thesis. These findings are summarized in Table 11.1 for passenger Hyperloop
system (Dtube = 2.23 m). While the alpha document only incorporated a 10% cost
margin, examining other large-scale infrastructure projects suggested it is more standard to add a 20% cost margin. This effect was implemented in the revised costs. It is
also assumed that Solar Panels and Batteries will cost the same as the alpha proposal.
The capital cost to acquire land is estimated to be three times the alpha cost. The
stacked bar graph in Figure 11.11 visually depicts the increases in capital costs for
each category. The revised total cost is suggested to be $16.84 billion USD versus the
original alpha total cost of $6 billion USD. Even though the revisions nearly triple
the alpha proposal they are still well beneath the $68.4 billion USD price tag for the
California High Speed Rail. The revised costs may still be conservative in some areas
while overcompensating in other areas.
Alpha Cost
(Billion USD)
Revised Cost
(Billion USD)
0.21
0.26
0.14
0.05
0.65
2.55
0.60
1.00
0.54
0.21
0.26
0.10
0.21
1.10
7.65
1.50
3.00
2.81
Total
6.00
16.84
130
16
14
Margin
12
Tunnel Construction
Pylon Construction
Tube Construction
Capsule
Propulsion
Station & Vacuum Pumps
Alpha Cost
Revised Cost
Figure 11.11: Capital cost comparison. Alpha total cost is $6 Billion USD vs. Revised
total cost of $16.84 Billion USD.
131
In the end, funds should be invested into this mode of transportation if the net
expected social benefit is both high enough to compensate for the infrastructure
capital and operating costs and greater than the next best alternative. This will
be strongly dependent on volume of traffic in the corridor, expected time savings,
average willingness to pay of potential users, the release of capacity in congested
roads, airports or rail lines, and the net reduction of external effects. Further studies
are needed to more accurately evaluate this issue.
132
Chapter 12
Conclusion
Elon Musk provided an innovative spark to transportation technologies when he released an alpha document for a new mode of transportation called the Hyperloop.
While the document provided a complete overview with clear goals for the system,
more thorough engineering analysis was required to assess its technical viability as
well as its economical reality. Each subsystem was broken down into its fundamental
parameters and governing physical principles. These were then analyzed to evaluate
their effect on both their corresponding subsystem and the overall Hyperloop performance. Several modifications were suggested to make the system feasible. Issues
downplayed by the alpha document were brought to attention and their impacts were
discussed. Economics, politics, and other human factors were also considered to complement the engineering analysis to gain a top-level perspective on this new mode of
transportation.
The most important physical parameter needed to achieve the high speeds desired
by the Hyperloop is the tubes pressure. The magnitude of the aerodynamic drag
needs to be reduced to minimize energy input and to allow the capsule to coast
for the majority of the journey. The previous chapters analysis on tube pressure
and overall trip time from Los Angeles to San Francisco showed that tube pressures
around 10,000 Pa would take nearly two hours, while lowering the tube pressure to
the alpha documents recommendation of 100 Pa would only take 36.35 minutes. The
aerodynamics and vacuum technology analyses suggested lowering the pressure below
40 Pa would produce diminishing returns. An upper bound on the tube pressure
can be established by a maximum permissible trip time. In order to make the trip
in less than 40 minutes the tube only needs to be evacuated to 2000 Pa. Target
tube pressure and desired pump-down times will dictate vacuum pump technology.
133
Keeping station interval distances small will help reduce pump-down times as well as
introduce redundancy for better leak detection and fixes. Higher vacuumizing speeds
are needed for airlock chambers.
Propulsion analysis acknowledged the effectiveness of the Double-Sided Linear
Induction Motor (DSLIM) to accelerate the Hyperloop capsule to full speed and
established the requirements for 1g acceleration. The Hyperloop aims to be selfpowering by utilizing regenerative braking and solar energy. Mounting solar panels
on top of the Hyperloops tube was determined to be more than sufficient to handle
the propulsion systems energy requirements, and the additional energy could be used
by other subsystems such as the vacuum pumps or even sold for a profit. Reducing
the operational costs due to energy will help keep ticket prices low. The capital costs
of propulsion are dependent on the energy and peak power requirements. Because
these costs are only a small percentage of the total capital costs, adding several reboost stations throughout the Hyperloop journey should be implemented to mitigate
elevation changes as well as allow for more active speed control in the suburban areas.
The air bearing suspension system is the largest technical challenge of the Hyperloop. Even at high speeds, the aerodynamic lift created by the air bearing skis
is not large enough to maintain vehicle levitation, creating the need for substantial
compressed air input. This can be achieved through a higher pressure of the compressed air or a larger mass flow rate. High mass flow rates risk choking the flow in
the small gap region and creating shock waves, while the higher magnitude of compressed air will require larger power and coolant requirements from the compressor
system. Furthermore, the small gap will create the need for tight tolerances for tube
manufacturing and assembly. A precise control system is essential to avoid collisions
with the tube wall and to mitigate the risk of adverse effects similar to pilot-induced
oscillations. Maglev is a proven technology seen operational in high speed train systems throughout the world and more consideration should given to using it for the
Hyperloops suspension. The risks of developing the air suspension system to a high
enough standard for public transportation may be greater than the extra costs of
using maglev. Due to ambiguous cost categories found in the existing maglev systems, the additional cost of implementing maglev in the Hyperloop was not accurately
evaluated to be able to make a recommendation. It is imperative a more complete
cost analysis is done on maglev technology before spending more time developing air
suspension technology.
A compressor mounted on the capsules front is needed to prevent the flow from
becoming choked at the Hyperloops high speeds. It was found a diffuser is needed
134
to ensure the flow does not become supersonic at the compressors inlet. The heat
released from the compressor system requires as larger water coolant mass flow rate
than anticipated by the alpha document. The increase in the required water mass
has effects that propagate through the entire system.
The ratio of the capsule size to tube size should be kept small in order reduce
viscous drag and prevent the flow from choking. The overall tube and capsule size
will drive total capital costs. It is suggested a larger system be built for more passenger
comfort. User studies should be conducted where participants are required to sit in
a mock capsule for the entire trip length. Ergonomic feedback will drive capsule size.
Although the technical aspects of the Hyperloop are challenging, it will be feasible
to build an operational system. This will require a substantial amount of prototyping
and design refinement. The recommendations made concerning the systems aerodynamics, propulsion, suspension, control, vacuum technology, compressor, route, and
civil infrastructure can be used as a more developed foundation to advance Hyperloop work. While technical solutions will be found to ensure high performance and
passenger satisfaction, it will be other aspects that have the potential to keep the
Hyperloop from leaving the concept stage. Economics and community pressures will
dictate the future of the Hyperloop in the Los Angeles/San Francisco corridor.
It was proposed for the route to run along I-5 to minimize land acquisition costs.
The lateral g-force analysis of following this route at the desired Hyperloop speeds was
shown to be problematic as lateral g-force exceeded 0.5g and accelerations changed
rapidly. Land will need to be acquired to smooth the track path and maintain passenger comfort. Although this is not predicted to be an issue in the middle section of the
journey, many obstacles can arise in the suburban area due to existing infrastructure
and community pressures. Alternative solutions can be considered, but ultimately
costs for acquiring land will be much higher than the alpha document estimated.
Civil infrastructure costs make up the largest percentage of total capital costs and
they were determined to be underestimated based on material cost analysis, review of
other large-scale infrastructure projects, custom pylon manufacturing requirements,
and higher tunneling costs. After a detailed revision, the estimated capital costs of
the Hyperloop were determined to be $16.84 billion dollars, nearly three times the
cost of the alpha proposed $6 billion. Still, this cost is well beneath the projected
$68.4 billion of the California High Speed Rail System. Some of the revised costs
may still be conservative because several technological components have not been
proven on a large scale, and at this early stage there are many other unknowns and
unaccountable factors that will increase final cost. Even if the Hyperloop goes over
135
136
Bibliography
[1] Luke Abaffy. Can Engineers Make Hyperloop a Reality?
Record, (September):24, 2013.
Engineering News-
[7] Y. Cai and S.S. Chen. Control of Maglev Suspension Systems. Journal of Vibration and Control, 2(3):349368, January 1996.
[8] California High-Speed Rail Authority. California High-Speed Rail Program Revised 2012 Business Plan. Technical Report April, 2012.
[9] California High-Speed Rail Authority. California High-Speed Rail Project Cost
Changes from 2009 Report to 2012 Business Plan Capital Cost Estimates. Technical Report April, 2012.
[10] California-Nevada Super Speed Train Commission and American Magline Group.
California-Nevada Interstate Maglev Project ( CNIMP ) A Guideway to the
Future. Technical report, American Magline Group, 2008.
[11] William D. Callister and David G Rethwisch. Fundamentals of Materials Science
and Engineering: An Integrated Approach. Wiley, Hoboken, NJ, 4 edition, 2012.
137
[12] Cornell University Global Labor Institute. Pipe dreams ? Jobs Gained, Jobs Lost
By the Construction of Keystone XL. Technical Report 5, Cornell University,
2011.
[13] Gines de Rus. The Economic Effects of High Speed Rail Investment, 2012.
[14] Sandeep Dhamaja. Power from Regenerative Braking. In Electrical Vehicle Battery Systems, chapter Electric V, pages 4368. Newnes: Butterworth-Heinemann,
Boston, 2002.
[15] J. Patrick Donohoe. Induction (Asynchronous) Machines, 2010.
[16] Mohamed M Elmadany and Zuhair S Abduljabbar. Vehicle System Dynamics : Linear Quadratic Gaussian Control of a Quarter-Car Suspension Linear
Quadratic Gaussian Control of a Quarter-Car Suspension. International Journal
of Vehicle Mechanics and Mobility, (September):3741, 2010.
[17] Keith A. Escoe. An Introduction to Engineering Mechanics of Piping. In Piping
and Pipeline Assessment Guide, chapter 2. Gulf Professional Publishing, Oxford,
1 edition, 2006.
[18] Bent Flyvbjerg, Mette Skamris Holm, and S ren Buhl. Cost Underestimation
in Public Works Projects: Error or Lie? Journal of the American Planning
Association, 68(3), 2002.
[19] R.L. Forgacs. Evacuated tube vehicles versus jet aircraft for high-speed transportation. Proceedings of the IEEE, 61(5):604616, 1973.
[20] Nick Garzilla. The Transportation Innovation Act, 2014.
[21] Bhag S. Guru and H
useyin R. Hizirolu. Linear Induction Motor. In Electric
Machinery and Transformers, pages 678685. Oxford University Press, 3 edition,
2001.
[22] William H. Heiser and David T. Pratt. Freestream Knudsen Number. In Hypersonic Airbreathing Propulsion, chapter 2. AIAA Educaation Series, Washington,
D.C., 1994.
[23] Joao P. Hespanha. Linear Systems Theory, volume 4. Princeton University Press,
Princeton, 2009.
138
[38] Daryl Oster, Masayuki Kumada, and Yaoping Zhang. Evacuated Tube Transport Technologies (ET3): a maximum value global transportation network for
passengers and cargo. Journal of Modern Transportation, 19(1):4250, 2011.
[39] Greg Paula. Evaluation EPACTs Impact. Mechanical Engineering Magazine,
(June), 1998.
[40] Pfieffer Vacuum. CombiLine WU Data Sheet. Technical report, Germany, 2009.
[41] Pfieffer Vacuum. Vacuum Technology Know How. Technical report, Germany,
2009.
[42] Don Plotkin and Simon Kim. Maglev Guideway Cost and Construction Schedule
Assessment. Journal of Transportation Engineering, (195), 1997.
[43] Robert W. Pratt. Pilot-induced oscillations (PIOs). In Flight Control Systems:
practical issues in design and implementation, chapter 4, pages 150167. TJ
International, Padstow, Cornwall, 2000.
[44] Pro SwissMetro. SwissMetro - Technology, 1995.
[45] Giuseppe Quaglia and Massimo Sorli. Air Suspension Dimensionless Analysis
and Design Procedure. Vehicle Systems Dynamics, 35(6), 2001.
[46] Road Traffic Technology. Boston Big Dig, Central Artery/Tunnel Project. Massacusetts - Road Traffic Technology, 2013.
[47] Sami Roinila. H Control Design of an Active Vehicle. PhD thesis, Tampere
University of Technology, 2011.
[48] R. M. Salter. The Very High Speed Transit System. Technical report, Rand
Corporation, 1972.
[49] Olivier Sename and Luc Dugard. Robust H-Infinity Control of Quarter-Car
Semi-Active Suspension.
[50] Xu Shuo, Zhou Ke, Luo Zhe, and Yu Fan. Study of PID Control for a SemiTrack Air-Cushion Vehicle Based on Power Consumption Optimization. Institute
of Electrical and Electronics Engineers, 2006.
[51] Siemens, ThyssenKrupp, and Transrapid International. High-Tech for Flying
on the Ground . Technical report, Transrapid International, Berlin, 2002.
140
[52] Hebertt Sira-Ramirez and Carlos A. Ibanez. The Control of the Hovercraft
System : A Flatness Based Approach. In International Conference on Control
Applications, Anchorage, Alaska, 2000. IEEE.
[53] Alexander J. Smits. A Physical Introduction to Fluid Mechanics. John Wiley,
Princeton, 2 edition, 2012.
[54] Sandeep Sovani. Bringing the Hyperloop One Step Closer to Reality Through
Simulation, 2013.
[55] R. L. Strykowsky, T. Brown, M. Cole Chrzanowski, P. Heitzenroeder, G. H. Neilson, Donald Rej, and M. Viola. Engineering Cost & Schedule Lessons Learned
On NCSX. Technical report, Princeton Plasma Physics Laboratory, Princeton,
2014.
[56] Antoni Szumanowski. Energy Efficiency of Regenerative Braking. In Hybrid
Electric Power Train Engineering and Technology - Modeling, Control, and Simulation, chapter 10, pages 381405. Engineering Science References, Hershey, PA,
2013.
[57] W. J. Thayer. Transfer Functions for MOOG Servovalves. Technical report,
MOOG Inc., East Aurora, NY, 1965.
[58] Transportation Research Board of the National Academies. Transit Capacity and
Quality of Service Manual. Federal Transit Administration, Washington, D.C.,
2 edition, 2003.
[59] U.S. Department of Transportation: Research and Innovative Technology Administration. Transportation Fatalities by Mode. Technical report, Bureau of
Transportation Statistics, 2012.
[60] U.S. Geological Survey. The Trans-Alaska Pipeline Survives the Quake - A
Triumph of Science and Engineering, 2003.
[61] D. M. Van Wie, F. T. Kwok, and R. F. Walsh. Starting Characteristics of
Supersonic Inlets. American Institute of Aeronautics and Astronautics, (July),
1996.
[62] Yao Ping Zhang. Explicit Formula for Estimating Aerodynamic Drag on Trains
Running in Evacuated Tube Transportation. Applied Mechanics and Materials,
307:156160, February 2013.
141
[63] Yao Ping Zhang and Daryl Oster. Study on Materials and Structure of the
Tube Wall of Evacuated Tube Transportation. Advanced Materials Research,
487:582587, March 2012.
[64] Yaoping Zhang. Vehicle scheme of Evacuated Tube Transportation in the future logistics system. 2010 The 2nd International Conference on Computer and
Automation Engineering (ICCAE), 5:190192, February 2010.
[65] Yaoping Zhang. Evacuated Tube Transportation Routes: Going Overheard, At
Ground Level, Shallow Underground or Deeply Underground?, 2013.
[66] Yaoping Zhang. Numerical simulation and analysis of aerodynamic drag on a
subsonic train in evacuated tube transportation. Journal of Modern Transportation, 20(1):4448, September 2013.
[67] Yaoping Zhang, Daryl Oster, Masayuki Kumada, Jianye Yu, and Shengshan Li.
Key vacuum technology issues to be solved in evacuated tube transportation.
Journal of Modern Transportation, 19(2):110113, September 2013.
[68] Yaoping Zhang, Jianyue Yu, Maoxing Shen, Daryl Oster, and Chen Chen. Affecting factors and numerical value calculation relating to vacuumizing time in
Evacuated Tube Transportation. Proceedings of 2011 6th International Forum
on Strategic Technology, 7:246249, August 2011.
[69] Yaoping Zhang, Jun Zhu, Baigang Mi, and Maoxing Shen. Reasonable Figure
of Subsonic Train Front and Rear in Evacuated Tube Transportation Based on
Aerodynamics Consideration. Proceedings of the 2nd International Conference
on Electronic and Mechanical Engineering and Information Technology (2012),
pages 734739, 2012.
[70] Y.P. Zhang, S.S. Li, and M.X. Wang. Main Vacuum Technical Issues of Evacuated Tube Transportation. Physics Procedia, 32:743747, January 2012.
142
Appendix A
Important System Parameters
The following important system parameters were derived from the Hyperloop alpha
document and were used throughout the analysis of the Hyperloops subsystems.
Capsule Parameters
mc = 15000 kg
Lcapsule = 31 m
Hcapsule = 1.1 m
Wcapsule = 1.35 m
143
Tube Parameters
T = 293.15 K
= 1.8369247e5
= 1.4
Rair = 287
(Adiabatic Index)
J
kg K
ptube = 100 P a
kg
ptube
= 0.001188579 3
=
Rair T
m
(Tube Pressure)
(Density of Air in Tube)
Dtube = 2.23 m
2
Dtube
SAtube =
= 3.9057 m2
2
ttube = 23.0 mm
Propulsion Parameters
dprop = 4 km
Tgap = 398.15 K
gap = 2.31849e5
pb = 9400 P a
kg
pb
gap =
= 0.08226199 3
Rair Tgap
m
(Bearing Pressure)
(Density of Air in Gap)
144
Wski = 0.9 m
= 0.05
(Angle of Attack)
Nski = 28
145