Академический Документы
Профессиональный Документы
Культура Документы
S UMMARY R EPORT
2014 - 2015
ii
Objectives MC
Described below are the objectives set and agreed by both the project team and supervisor:
1. To design, build and test a 10¼´´gauge locomotive with an energy recovery system to recover
kinetic energy otherwise lost during braking
2. To successfully plan and deliver the project on time and in budget, with appropriate documen-
tation
3. Where possible obtain parts and or sponsorship from companies (local or otherwise) in order
to maximise awareness of the project and reduce expenditure
4. To design said locomotive such as to comply with the IMechE Railway Challenge technical
guidelines and rules
5. Entry into the IMechE Railway Challenge competition in June 2015
6. To visit and present the project to potential sponsoring companies
7. To produce promotional material for publicity/recruitment/industrial engagement for use by
the faculty
8. To develop and deliver a comprehensive business case that will convince the executives of
an assumed large corporation that the proposed design best meets their demands.
Resources MC
• Total budget: £23,260.00
Constraints MC
During the project the team overcame a number of difficulties including but not limited to:
• Project management
• Bogies
• Brakes
• Electronics & control
• Energy recovery
• Frame
• Shell
• Traction.
Team Organisation MC
Student team members:
Mitchell Clark Project management
Qasim Al Mansoor, Matt Marinaccio, Jack White Electronics and control
Callum Livingstone, Hamish Munro-Faure, Peter Nixon Energy recovery
iv
Academic support:
Important Results MC
Based on the combined efforts of all team members, the project has successfully delivered a func-
tional miniature locomotive (Figures 1 and 2) with the means to recover energy from the braking
process. A short summary of the final design follows (further details can be found in the respec-
tive technical chapters or a general summary in Section 12). The locomotive has been manufac-
tured using aluminium frame and incorporates two bogies (Figure 3), each with two chain-driven
wheelsets. A failsafe vacuum brake system has been implemented along with two DC Lynch mo-
tors in a monomotor arrangement. A supercapacitor has been used to store recovered braking
energy and primary power is provided by a petrol generator. An aluminium plate shell will provide
weatherproofing for the system whilst a myRIO mictrocontroller provides the means to control the
entire system via a Wi-Fi linked tablet.
As well as the physical locomotive the team are able to hand over the mechanical design CAD
files & a respective part index tracking all of the aforementioned parts. A comprehensive bill of
materials for the locomotive has also been updated throughout the project allowing the origins of
any parts used to be quickly traced back to a supplier or vendor if required.
Conclusions MC
A 10¼" gauge miniature locomotive incorporating a number of subsystems has been successfully
designed, manufactured, assembled and tested to comply with a technical specification as provided
by the IMechE for the Railway Challenge competition. As one of, if not the largest GDP teams, the
team have successfully worked together in a limited time frame to construct a functional locomotive
in its entirety. Further work has been identified and recommended both for activities post-report
hand-in in preparation for the Railway Challenge competition by the current team and also for next
year’s project team. The efforts of the current team have resulted in a flexible initial design such
as to allow subsequent teams to optimise the existing design and control systems. Infrastructure
such as the means to safely raise the locomotive to an ergonomic height for construction (build
frame), ability to move the locomotive once at floor level (jacking system) and track panels to allow
local testing of the locomotive have all been designed or acquired as needed. Useful contacts
and connections have been forged both at university by way of specialist professors or lecturers
who can assist in future years, the local miniature railway Eastleigh Lakeside Railway for testing
and advice, the IMechE for subsequent competition entries and suppliers/manufacturers such as
Clarke Lane Engineering based very close to Highfield Campus. As well as attracting attention
v
across the whole of the engineering faculty, the project has drawn particular interest from the Dean
of Engineering, Professor William Powrie, been used for a science and engineering open day in
collusion with the Track 21 project and will be shown at the inaugural faculty design show in June.
Finally, the team successfully applied for and won a "rookie team grant" from the IMechE (as a
first-time team entrant into the Railway Challenge) to the value of £1,500 which is to be used in
addition to the basic GDP budget for next year’s team.
Recommendations MC
As the current project is the first of its kind in a considerable number of years, efforts were con-
centrated on the design, manufacture and assembly of a functional locomotive and associated
infrastructure such as the build frame. As a result, future teams will be able to focus more on
optimising the existing locomotive by making (assumed) minor changes to the mechanical and
electrical system designs proposed in this report. Particular areas the team feel could be improved
upon are as follows:
• Redesign of the bearing adapters located between the bogie frame and the wheelsets. These
parts were submitted to the EDMC very early on in the project and, as such, were retrospec-
tively modified as their purpose evolved with the project. The opportunity for future teams to
simplify and optimise their design exists and as such is a suggested area for future work.
• Optimisation of the ancillary battery charging circuit through implementation of algorithms to
improve the efficiency of battery charge. In particular, the use of the myRIO microcontroller
to control the battery charging.
• Further work to isolate generator vibrations through the frame and rest of the locomotive
• Possibility of using a fuel cell as the primary energy source could be considered. This would
reduce the vibration in the locomotive and provide a source of innovation which the current
team did not use (though did consider).
• Further work recommendations with respect to the IMechE Railway Challenge will largely
depend on the changes to the technical specification, technical guidelines and or the changes
to the competition challenges in the 2016 competition which have not yet been released.
vii
M AIN R EPORT
2014 - 2015
Contents
Page
List of Figures 5
List of Tables 9
Nomenclature 10
1 Introduction 13
1.1 A Brief Introduction to UK Railways . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.2 Miniature Railways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3 Eurostar Project - Southampton University . . . . . . . . . . . . . . . . . . . . . . . . 14
1.4 IMechE Railway Challenge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4 Traction 23
4.1 Traction Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.2 Initial Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.3 Required Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.4 Overall Traction System Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.5 Electric DC Motor Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.6 Drivetrain and Transmission Selection/Design . . . . . . . . . . . . . . . . . . . . . . 34
4.7 Generator Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.8 Motor Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.9 Design and Manufacture of Traction System . . . . . . . . . . . . . . . . . . . . . . . 39
5 Energy Recovery 41
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.2 Potential Energy Recovery Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.3 Supercapacitors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.4 Energy Recovery Challenge: Initial Model . . . . . . . . . . . . . . . . . . . . . . . . 51
5.5 Electrical Power Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.6 Component Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.7 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.8 Energy Recovery System Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6 Brakes 74
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.2 IMechE Specification for Brakes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.3 Braking Force - Initial Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.4 Research into Full-Size Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.5 Potential Systems for Miniature Locomotive Design . . . . . . . . . . . . . . . . . . . 76
6.6 Proposed Designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.7 Integration with Other Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.8 Final Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.9 Final Calculations for Braking System . . . . . . . . . . . . . . . . . . . . . . . . . . 84
2
CONTENTS 3
13 Testing 190
13.1 Jacking Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
13.2 Locomotive Test Run . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
15 Conclusions 197
Bibliography 198
4 CONTENTS
Appendices 208
A Bill of Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
B Risk Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
C Traction Data Sheets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
D Supercapacitor Data Sheets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
E Battery Data Sheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
F EDMC Drawings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
G Drawbar Drawing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
H Bosch Beam Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
I Additional Suspension Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
J Bogie Curving Properties Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
List of Figures
9 Summary of primary fuels/power sources and their relationships with various types
of traction motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
10 Forces acting on the locomotive while it is in motion . . . . . . . . . . . . . . . . . . 27
11 Excel spreadsheet showing performance calculations for specification of traction
motors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
12 Excel spreadsheet showing performance calculations for specification of generator . 32
13 Various DC traction motors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
14 Lynch motors being bench tested . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
15 Example of an epicyclic gear set [21] . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
16 CAD rendering of the proposed transmission design . . . . . . . . . . . . . . . . . . 36
17 Two possible DC generator sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
18 Generator and SMPS chosen to power locomotive traction system . . . . . . . . . . 39
19 Motor mounting and FEA results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
20 Sprocket adapter and FEA results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
21 Motor with machined spacers, small sprockets and mount . . . . . . . . . . . . . . . 40
5
6 LIST OF FIGURES
35 The change in voltage of the supercapacitor with respect to the time taken to re-
generate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
36 The velocity of the train under energy gained from regenerative braking . . . . . . . 69
34 Circuit diagram detailing the arrangement of components used to simulate the lo-
comotive energy recovery system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
37 The distance covered by the train using previously recovered energy . . . . . . . . . 71
38 Circuit diagram of the energy recovery bench top test . . . . . . . . . . . . . . . . . 72
39 Supercapacitor charge and discharge rates . . . . . . . . . . . . . . . . . . . . . . . 73
50 Different microprocessors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
51 Binary weighting matrix showing the relative importance of microprocessor param-
eters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
52 Overall score of microprocessors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
53 NI Data Dashboard application [91] . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
54 Schematic arrangement of optical speed measurement [92] . . . . . . . . . . . . . . 92
55 Speed sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
56 LabVIEW RPM VI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
57 Speed sensor test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
58 MATLAB code for RPM timer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
59 Speed sensor and magnet brackets . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
60 Potential divider circuit for temperature measurement . . . . . . . . . . . . . . . . . . 97
61 Fire detection circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
62 LabVIEW temperature measurement VI . . . . . . . . . . . . . . . . . . . . . . . . . 98
63 Upstream and downstream pressure against volumetric flow rate [96] . . . . . . . . 101
64 Circuit diagram for the standard transistor switching circuit . . . . . . . . . . . . . . . 101
65 Initial VI for all switching circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
66 Motor control binary weighting matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
67 Motor control ancillary circuitry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
68 Illustration of four quadrant motor control operation [97] . . . . . . . . . . . . . . . . 106
69 Braking system design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
70 Motor control bench testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
71 Battery monitoring components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
72 Full 12 V ancillary circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
73 Low voltage disconnect circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
74 Voltage divider test circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
75 User interface of voltage monitoring NI LabVIEW script . . . . . . . . . . . . . . . . . 114
76 Block diargram script that monitors the battery voltage via the myRIO analogue input 115
77 Theoretical current sensor circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
LIST OF FIGURES 7
113 von Mises FEA stress plot for steel frame design . . . . . . . . . . . . . . . . . . . . 157
114 View of the frame showing the lateral generator support beams fixed on top of the
main longitudinal beams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
115 Experimenting without supporting beams . . . . . . . . . . . . . . . . . . . . . . . . 159
116 Initial and final superstructure designs . . . . . . . . . . . . . . . . . . . . . . . . . . 159
8 LIST OF FIGURES
117 von Mises FEA stress plot for 5 mm aluminium shelf design . . . . . . . . . . . . . . 160
118 Contactors being attached to the shelf by the E&C subteam . . . . . . . . . . . . . . 160
119 Early design using double gusset arrangement . . . . . . . . . . . . . . . . . . . . . 161
120 Bosch connector load ratings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
121 View of the headstock beam showing the 45×45mm connectors attached between
the longitudinal beams and the headstock . . . . . . . . . . . . . . . . . . . . . . . . 161
122 Double bogie beam arrangement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
123 Use of S12 connection screws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
124 Initial coupler designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
125 Von Mises FEA stress plot for coupler design . . . . . . . . . . . . . . . . . . . . . . 164
126 Jacking designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
128 Final frame geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
127 Frame FEA study for locomotive at rest with 1500 N backward-acting force to simu-
late maximum acceleration up a 2.86° slope with a 600 kg trailing load . . . . . . . . 168
5 Binary weighting matrix showing relative importance of energy storage medium pa-
rameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
6 Overall score of energy storage medium suitability . . . . . . . . . . . . . . . . . . . 47
7 Assumptions of the initial model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
8 Power required by the train in different operating situations . . . . . . . . . . . . . . . 54
9 Switch ratings and functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
10 Calculated estimates of system component inertia values . . . . . . . . . . . . . . . 64
11 Lynch motor parameter inputs for Simscape simulation . . . . . . . . . . . . . . . . . 64
12 Supercapacitor parameters input for Simscape simulation . . . . . . . . . . . . . . . 65
13 Calculations made to assess required force output from any locomotive braking
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
14 Binary weighting matrix showing the relative importance of braking parameters . . . 79
15 Brake system calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
20 Typical values for each parameter that appear in Equation 8.5.7. . . . . . . . . . . . 136
21 Critical frequencies and SPLs at the receiver for various shells . . . . . . . . . . . . 173
22 Advantages and disadvantages of aluminium and fibreglass . . . . . . . . . . . . . . 174
9
Nomenclature
10
LIST OF TABLES 11
The team wish to acknowledge the following for their assistance during the project:
Dr Mohamed Torbati - Project Supervisor for supervising the project
Professor Sulemain Sharkh - Mechatronic Project Advisor for mechatronic project advice and
guidance
Professor David Thompson - Rail Vibration Project Advisor for advice and guidance regarding
bogie and suspension design
Mr Vorrapath Kokaew - Simulink Project Advisor for initial advice and guidance with use of
Simulink
Dr John Atkinson - Second Assessor for acting as examiner for the project.
Tim Hartley - Lab Technician for above average chat, invitations to Chef Hong Kong for lunches
and copious amounts of invaluable life advice
Pavittar Bassi - Purchasing (Engineering) for efficiently dealing with as many order requests
from this GDP as the rest of the projects combined
Siddiq Al-Busmait - Previous IMechE Railway Challenge Winner for technical advice and first
time project entry help
Alan Stewart - Siemens for project management guidance, site visits, technical advice and indus-
trial sponsorship (although we really would have liked that toolbox)
Paul Darling - Siemens for assistance with items purchased through Siemens
Beejal Shah - National Instruments for initial myRio training and technical advice
Clive Upton and team - Eastleigh Lakeside Railway for helpful advice & guidance and for allow-
ing the use of the track for testing purposes and to gather data from
Dr Martin Toward - ISVR for helping during suspension testing and providing all the necessary
equipment
John Young - EDMC for assisting the team in getting the wheelsets machined in time for the Uni-
versity of Southampton science fair
Justin Gregory - EDMC for almost daily delivery notifications and enough nuts & bolts to build a
train
Rachel Pearson - IMechE for swift rule clarification where required
Gary Painting & Team - Estates and Facilities for transporting the locomotive to Eastleigh Lake-
side Miniature Railway from Highfield Campus for testing
Cliff Terry - IMechE for assistance and clarification in preparation for the Railway Challenge
IMechE for awarding the team with a £1,500 rookie team grant which will be used by next year’s
team
Rob Stansbridge - ISVR for invaluable electronics and sensing systems advice.
12
1. Introduction
1.1 A Brief Introduction to UK Railways MC
Historically, the railway industry has always had a strong association with Great Britain. Since the
world’s first steam-hauled railway running between Liverpool and Manchester opened in 1830 as a
replacement for horses, carts and canals, Great Britain’s railway industry has surged in size [1]. The
total number of journeys by rail in Great Britain more than doubled between 1994/95 and 2009/10.
It is predicted that the average person will make 61% more long distance rail journeys by 2043 in
comparison to current levels [2]. In light of these trends and predictions, UK rail infrastructure is
undergoing heavy investment from the government in order to reduce journey times and make rail
travel a more attractive prospect for potential customers. The two most significant projects currently
being undertaken are those of Crossrail in London and High Speed Two (HS2), the latter of which
is currently estimated to cost around £42.6bn and is due for completion in 2033 [3].
The motivations of the government to invest in the railway are not just profit driven, but also
based upon emissions and energy usage. Due to the potential number of passengers transportable
by a single carriage, trains are able to overcome their high speeds and heavy frames by greatly
reducing the frontal area per passenger. This results in trains having a hugely reduced drag force
per person in comparison to a car with a single occupant. Trains also only ever have a very small
contact region between wheel and rail which results in minimal rolling resistance. The outcome
of these facts is that, per person, trains are much more efficient in energy terms than personal
vehicles. A commuter train is upwards of 25 times more efficient than a standard road car [4].
A current area of interest for the rail industry is that of energy recovery. That is, recovering
energy otherwise lost during the braking process and storing it somehow for use as tractive effort
to improve efficiencies. This can take the form of local storage such as a supercapacitor/flywheel
or in grid form whereby recovered energy is sent back to the supply grid for use elsewhere.
13
14 CHAPTER 1. INTRODUCTION
(a) 10¼" gauge rail at Stapleford Miniature (b) Selection of 10¼" gauge locomotives at
Railway in Leicestershire Eastleigh Lakeside Miniature Railway
railway is a dual-gauge variety with over 1.25 miles of both 7¼" and 10¼" tracks. A fleet of 19
steam, one battery and two diesel hydraulic powered locomotives are located at the miniature rail-
way [6]. The battery powered locomotive is the Eurostar, a previous University of Southampton MSc
project, which has since been modified and improved by Eastleigh enthusiasts. See Section 1.3
for further information about this particular locomotive. The railway enthusiasts at Eastleigh were
incredibly useful in the early stages of the project, allowing team members full access to the railway
and locomotives. This provided a sound opportunity for the group to see the subsystems and mech-
anisms in use on such miniature locomotives, as well as gratefully receive technical advice from the
incredibly experienced volunteers. During the latter parts of the project, the team were given open
access to the railway to allow the testing of the assembled locomotive, see Section 13.2 for further
details. The close locality of Eastleigh Railway to the University is a superbly useful resource and
future teams are strongly urged to make full use of the incredibly generous and knowledgeable
volunteers based there, as well as the obvious advantage of a working track (with both bends and
points!).
under the supervision of Professor Sharkh. The four man project was based at Eastleigh Lakeside
Railway where they were able to build and test the locomotive on site. The Eurostar now perma-
nently resides at Eastleigh Railway where it is maintained by enthusiasts and remains in frequent
use. In the early stages of the current project (as mentioned above), some of the team visited
Eastleigh to inspect and look at both the track and the previous Eurostar project.
The Eurostar miniature locomotive, see Figure 6a, is around 3 m in length and is of 10¼" gauge,
like that of the locomotive which has been designed. Outlined below are the various design aspects
of the Eurostar including feedback from volunteers at Eastleigh Railway:
• Bogies The Eurostar has two bogies, each with two driven wheelsets (see Figure 6c). The
wheelsets are constructed from EN8, a variety of medium carbon steel, and each axle has
a brake disc welded to it (see Figure 6b). The bogies interact with the frame by means of a
simple plain bearing (a lubricated shaft in a slightly larger hole), with lubricated nylon blocks
at either side for alignment purposes. The locomotive has only single suspension with coil
springs mounted between the axlebox and bogie, damping vibrations from discontinuities in
the track.
• Brakes The Eurostar makes use of a vacuum braking system whereby a brake cylinder actu-
ates an external rod based on the movement of internal diaphragms due to a vacuum formed
in the system. A vacuum pump is mounted centrally on the frame. A single cylinder per bogie
is linked to a calliper which grips a brake disc mounted on each axle.
• Electronics & control The most notable electronics and control aspect of the Eurostar is the
sizeable battery pack mounted in the centre of the frame (Figure 6d), allowing the locomotive
to run for multiple days from a single charge. The control panel (Figure 6e) for the locomotive
is simple with controls for throttle, direction, brake application, horn and lights. There are also
indicators for battery condition, vacuum level for the brakes, wheelslip and a modified bicycle
computer for speed.
• Frame The chassis is constructed of tubular steel which provides mass to aid in tractive force
to reduce wheelspin under power. The size of the Eurostar is largely due to the size of the
required battery pack mounted centrally on the frame.
• Shell The Eurostar shell is a mixture of fibreglass and wood construction with a completely
removeable box section (Figure 6f) which has a hinged access door. As the Eurostar is a
standard miniature railway locomotive there is also a suitable cut-out in the exterior for a
driver to sit in.
• Traction The Eurostar is powered by two brushless DC motors, one mounted on each bogie.
All wheelsets are chain driven by their respective motor. The volunteers noted that one of
the first modifications undertaken when they first received the Eurostar was to remove the
previously installed bicycle chain drive and swap it for a heavier duty motorcycle chain.
16 CHAPTER 1. INTRODUCTION
(a) Front view of Eurostar locomotive showing fi- (b) Carriage wheelset showing primary suspen-
breglass nose sion, brake disc and calliper
(c) Top-down view of bogie showing motor, chain (d) Eurostar with shell removed showing battery
drives and wheelsets pack
(e) Close-up view of the driver’s cab in the Eu- (f) Wooden Eurostar shell when removed, access
rostar hatch opened
• Energy efficiency whereby the primary energy utilisation of the locomotive is assessed
against the useful energy output transferred to the wheels
• Energy recovery/storage whereby the locomotive must come to a stop by a given point and,
during this time, recover as much energy as possible. Once stopped, the locomotive must
utilise the stored energy for tractive effort in an attempt to propel the locomotive as far as
possible using only recovered energy.
• Traction whereby the ability of the locomotive to pull a heavy trailing load up a 5% gradient
is assessed
• Ride comfort where the aim is to reduce the vibration transmitted to the outside of the loco-
motive as much as possible so as to improve the theoretical ride comfort
• Noise simply to reduce the noise output of the locomotive
• Maintainability where the aim is to consider the maintainability of the locomotive such that a
wheelset can quickly be swapped out in the fastest time possible
• Design where the design considerations of the locomotive are investigated including meeting
specified requirements and innovative design aspects
• Business case whereby a theoretical pitch to a large corporation who are considering the
purchase of the designed locomotive must be presented to a judging panel considering as-
pects such as environmental impact and profitability.
Teams must also submit technical documentation, including a summary report and proof of calcu-
lations, to justify design decisions made and to prove the locomotive is safe for testing.
One of the principal motivations behind the project is that the locomotive be designed with the
competition in mind such that it can be entered into the competition in June 2015.
2. Project Brief & Aims MC
1. To design, build and test a 10¼" gauge locomotive with an energy recovery system to recover
kinetic energy otherwise lost during braking
2. To successfully plan and deliver the project on time and in budget, with appropriate documen-
tation
3. Where possible, to obtain parts and or sponsorship from companies (local or otherwise) in
order to maximise awareness and reduce expenditure
4. To design said locomotive such as to comply with the IMechE Railway Challenge technical
guidelines and rules
5. To enter into the IMechE Railway Challenge competition
6. To visit and present the project to potential sponsoring companies
7. To produce promotional material for publicity/recruitment/industrial engagement, used by the
faculty
8. To develop and deliver a comprehensive business case that will convince the executives of
an assumed large corporation that the proposed design best meets their demands.
18
3. Project Overview & Budgeting
3.1 Team Structure & Task Allocation MC
Team members found out about their allotted group project late in the 2013/14 academic year,
around April 2014. With the project assigned the group initially met in May/June 2014 to discuss
the project and try to divide the work into discrete areas ready for assignment in September 2014
after the summer break.
The initially proposed breakdown of work by the project supervisor was to split the work into
areas such as braking & energy recovery, mechanical design, noise, performance & structural cal-
culations and traction control. However, in early meetings it was decided by the team to redistribute
the work according to subsystems such as to more appropriately break the work down into discrete
work packages for each subsystem. The chosen work packages and the allocation of responsible
team members were as follows (Figure 7):
As indicated in Figure 7 the team consisted not only of students but also academics and Siemens
as an industrial sponsor. Principal academic support was as follows:
19
20 CHAPTER 3. PROJECT OVERVIEW & BUDGETING
Before the project started it had been agreed with Siemens that they would sponsor the project
to the sum of £5,000 financial sponsorship and also provide technical assistance where required
(see Section 3.2 for further financial details). The team were also able to visit Eastleigh Works,
a site where South West Trains locomotives, which Siemens maintain, undergo heavy overhaul.
This provided a useful opportunity for group members to see commercial locomotive bogies up
close and gain an appreciation for aspects of design which ease maintainability of locomotives, as
required for the IMechE competition.
3.2 Budgeting MC
As the project was a complete design, build and test proposal starting from no real previous work,
the anticipated costs were considerable. The initially allocated budget at the beginning of the
project was as follows:
During the early months of the project, the team were fortunate enough to receive £17,000 of
the £22,750 grant money from an Engineering faculty grant fund which Dr Torbati applied for in
anticipation of the high project costs. This meant the operating project budget was increased as
shown below:
An initial budget split occurred after a few weeks of research by the individual subteams. This
initial split was then revised close to the project review in week 11 of semester 1, somewhere
around halfway through the project. Table 1 shows the initial and revised budget splits for the
project. Figure 8 shows the final project spend between the various subsystems. It is observed
that the major differences between predicted and final spend come from the electronics & control
and traction subsystems.
3.3. APPROACH TO THE PROJECT 21
A final bill of materials (BOM) for the entire project spend can be found in Appendix A.
of the project, the whole team met once a week. During these meetings a round-the-table discus-
sion on the progress made that week occurred as well as providing a floor for discussion about
design decisions and the opportunity to voice any concerns or difficulties encountered. After a
few weeks, weekly summary reports as prepared by Mitchell provided a basis for the discussion in
these meetings such as to add to the structure of the discussion. Achievements for the week were
obtained from individual group members in advance of the group meetings and compiled into a
document which was then displayed during the meetings. Progress against a project plan was also
assessed to allow any upcoming important milestones to be highlighted and also ensure progress
was occurring in a timely fashion.
Subteams with multiple members met as often as required, particular collaboration was required
between the electronics & control subteam and the other subsystems to be controlled such as
brakes, energy recovery and traction.
Once initial research had been undertaken, subsystems began to require various parts to be
purchased such that modelling and or testing could occur. All order requests were submitted to
purchasing by Mitchell, thus allowing careful control of the overall spend at all times (a problem
which could otherwise easily have been encountered with such a large group).
An initial report structure was created in late 2014 using LATEXby Mitchell, using the Eurostar
report as a basis for the division of the sections. Team members were then able to add to this as
they saw fit throughout the duration of the project. Project review and VIVA presentations were/will
be created in a similar fashion with an initial template/layout created and then team members
editing their respective sections to add content. This ensured consistency in both the layout and
style of such documentation.
A copy of the project risk assessment can be found in Appendix B.
4. Traction
4.1 Traction Overview HM
The traction system, which incorporates the primary power generation, transmission and drivetrain,
is what makes the locomotive move. In the same way a car has an internal combustion engine,
gearbox and drive shafts, the locomotive needs a way to convert the stored energy in a primary
fuel or power source into torque through the axles and wheels to provide a tractive driving force.
The number of different primary fuels, methods to convert their stored energy into motion, types
of transmission and options for the drivetrain led to a significant amount of initial research being
performed to determine an overall concept system, for which individual components could then be
either selected or designed. This initial research is covered in Section 4.2, with details of the selec-
tion and design of the individual components in Sections 4.5, 4.6 and 4.7. Determining the required
performance of the locomotive and gaining an insight into how the final design would theoretically
perform was critical to enable specification of the design and selection of the components, and
details of this process can be found in Section 4.3. Although all of the research is presented as a
step-by-step process, finding an overall system concept that would be both effective and feasible
relied heavily on design decisions made by the whole team and went through a number of iterations
before the concept discussed in Section 4.4 was chosen.
Detail on the design and manufacture of the whole traction system is found in Section 4.9. The
design work ranged from individual components to incorporating the entire system into the loco-
motive. Manufacture of certain bespoke components was carried out using the facilities available
at the University in the EDMC.
23
24 CHAPTER 4. TRACTION
flow and pressure to create torque and movement. Thirdly pneumatic motors were found, similar to
hydraulic but using compressed air flow rather than liquid. The final option considered was using a
direct mechanical linkage from an internal combustion engine (ICE) to drive the wheel sets.
To compare the four choices, a number of criteria were outlined to make a decision, including
the following:
Although these chosen criteria were broad they served as an easy way to make an initial de-
cision on which type of traction motor would be used. As the electric motor is the standard used
in nearly all modern locomotives, the initial research was mainly to see if any of the other options
had any tangible benefits over what is already established. Table 2 summarises the findings of the
comparisons, showing the general advantages and disadvantages of each system.
Table 2: Comparison of advantages and disadvantages of different types of traction motor [10] [11]
[12]
4.2. INITIAL RESEARCH 25
As well as the advantages and disadvantages highlighted according to the criteria, there was
also research into the possible primary fuel and power source options for each traction motor
type. This can be seen next in Section 4.2.2. The decision on which traction motor and primary
fuel/power source is detailed with reasoning in Section 4.2.3.
Stored Compressed
Primary Fuel Type Electric Battery Hydrogen Fuel Cell Petroleum Based
Air
Internal
Traction Method Electric Motor Hydraulic Motor Pneumatic Motor
Combustion Engine
Figure 9: Summary of primary fuels/power sources and their relationships with various types of
traction motor
26 CHAPTER 4. TRACTION
Twheel Key
= Locomotive
Fwheel = Trailing Load
Mwheel.g.cos(α) = Weight
= Driving Force
= Parasitic Force
FRR = Normal Force
Fnormal = Torque
FTraction
FDrag
At first this appears to be a simple mechanics problem; a mass on a slope with a number of
forces acting on it. However, it can also be seen that the wheel to rail interface is more complex, with
frictional forces determining the maximum tangential driving force, and therefore supplied torque,
that each wheel can provide. The resultant forces on the locomotive can be summarised as seen
in Equation 4.3.1:
X
F = FTraction − FDrag − Mloco g sin(α) − Mextra g sin(α) (4.3.1)
Where FTraction is the total tractive force driving the locomotive forward, FDrag is the combination of
rolling resistance and aerodynamic drag, Mloco and Mextra are the mass of the locomotive and the
trailing load respectively, g is acceleration due to gravity and α is the angle of the slope.
If we let N be the number of driven wheels in contact with the rail, Mwheel be the mass of the
locomotive acting through each wheel and CF be the wheel to rail friction coefficient, then the
maximum tangential driving force for each wheel, Fwheel , can be defined as seen in Equation 4.3.3:
This gives us the maximum tractive force that the wheel and rail interface can provide without wheel
28 CHAPTER 4. TRACTION
spin for any given CF value. This is a useful value to obtain as the traction event in the IMechE
Railway Challenge is scored on how quickly the locomotive can accelerate from standing and travel
a certain distance while pulling the trailing load, meaning that optimum tractive effort with as little
wheel spin as possible during acceleration is sought after.
Letting CRR be the coefficient of rolling resistance, then the resistive force on each wheel can
also be calculated as seen in Equation 4.3.5:
However, FDrag is made up of both the rolling resistance of the wheels on the track and also the
aerodynamic drag on the train. To find this, the equation to calculate aerodynamic drag force, FD ,
was used [14]. This can be seen in Equation 4.3.6:
Where CD is the coefficient of drag, Afrontal is the frontal area, ρair is the mass density of air and u
is the forward velocity. From this, the total resistive force on the locomotive, FDrag , can be defined
as seen in Equation 4.3.7:
Using Equations 4.3.1, 4.3.4 and 4.3.7, a model of the locomotive could be developed with optimal
times for acceleration and performance. Parameters such as mass, number of driven wheels and
rail/wheel friction coefficient were adjustable, allowing the team to see the effects that changing
them caused.
However, to use the model to specify the performance requirements of the electric motors,
drivetrain and generator, further calculations were required. The values of maximum tangential
driving force per wheel, Fwheel , at different rail/wheel friction coefficients, can be used to find the
maximum torque per driven axle, Taxle , where each axle drives two wheels. This can be seen in
Equation 4.3.8:
Taxle = 2 × Fwheel × Rwheel (4.3.8)
Where Rwheel is the radius of the wheels. This relationship allowed the drivetrain and motor to be
designed to be able to provide adequate torque for full speed acceleration at the expected rail/wheel
friction levels. The required torque and speed of the motor to achieve this full speed acceleration
gave a required power input. This then allowed for the generator to be sized accordingly.
With the theory and governing equations established, certain parameters were fixed before any
calculations were made. The justification for these parameters being set as they are can be seen
in Section 4.3.2, with the calculations and results found in Section 4.3.3.
Table 3: Parameters affecting traction performance stated in IMechE Technical Specification [15]
These values gave a baseline to develop the traction system from. Of particular note were the
maximum operating speed, range of friction coefficients and the trailing load mass. The wheel size
for the locomotive was decided as Rwheel = 125 mm based on calculations performed in cooperation
with the Bogies sub-team which can be seen in Section 8.9.1. It was also decided that there would
be a total of four axles and that ideally all eight wheels would be driven to maximise the possible
tractive effort, meaning that N = 8.
To find a suitable coefficient of rolling resistance, research was done on typical values found
in the rail industry. Generally, rolling resistance increased as axle load and travelling speed de-
creased. Figures for full sized locomotives are in the range of CRR = 0.00013 - 0.0002 [16]. How-
ever, as the miniature locomotive has a significantly lower axle load and speed than a full sized
equivalent, a value of CRR = 0.004 was chosen as an estimate, which, although likely too high,
will have a small effect on the performance calculations due to the low value of rolling resistance
with steel wheel on steel rail contacts anyway. The efficiency of the motor and drivetrain is as-
sumed to be 85%, which reflects the high efficiency of most electric motors and most mechanical
transmissions.
The information given so far allows calculations on the required axle torque and speed, and
therefore motor and drivetrain specification, needed for optimum locomotive performance in the
traction event. However, to be able to specify the generator and get an estimate of the total energy
that will be required to run the locomotive, the guidance in the IMechE Technical Specification was
used, that states“The locomotive shall provide self-contained motive power and be able to operate
for three hours without refuelling, on the basis that operation comprises continuous travel at 5 km/h
on a 5% gradient with a 400 kg trailing load of negligible friction” [15]. This information provided
sufficient input parameters to get an idea of the total energy capacity that would be required.
idea of the motor specifications that would be required for the locomotive. The spreadsheet used
can be seen overleaf in Figure 11.
Taking an average friction coefficient of CF = 0.35, it can be seen that the power requirement for
maximum acceleration is around 3.75 kW per motor. The speed and torque, which multiply to give
the power requirement, can be adjusted through the gearing ratio, giving a wider range of possible
motor options. For this friction coefficient, a predicted acceleration time from standing to full speed
is 2.87 s while attached to the trailing load, assuming no gradient. Two motors in the 3.5 kW to
6 kW rated power range were decided as suitable for the locomotive.
For specifying the generator, and the energy capacity required for operation, further calculations
were carried out. These took the assumed average friction coefficient of CF = 0.35 and used the
IMechE guidelines on typical operation. The spreadsheet for these calculations can be seen in
Figure 12.
Total energy consumption of around 2.6 kWh was found to be reasonable. This could be
achieved with a generator only producing around 1 kW of power for a number of hours, but this
was found to likely limit the available current and torque. Therefore, a 2.6 kW - 3.0 kW generator
capable of a run time of 3 hours or more under normal load was decided to be the best option.
With these results, an overall system concept could be specified, as seen in Section 4.4.
Due to the fact the rest of the system would be largely DC, and that AC motors were often
more expensive, it was decided to discount them and focus on choosing DC motors that would
meet the requirements. The relatively simple speed control and high levels of experience available
to the team from the faculty also made the case for DC motors to be chosen. With the concept
specified, the components to satisfy the requirements needed to be chosen and designed for,
which is covered in the next five sections.
Traction Motor and Transmission Requirements Calculations Sheet 1: Mechanical Calculations for Finding Required Motor/Gearing Specifications
Motor/Drivetrain Calculations
Wheel/Track Max. Wheel T_axle: Max. Axle Torque Req. Torque Motor Torque Req. Motor Req. Motor Req. Motor
Friction Coeff. Drive Force (N) (Nm) Out of Trans. (Nm) w/ G:1 Ratio (Nm) Speed w/ G:1 (rad/s) Speed w/ G:1 (rpm) Power w/G:1 (W)
0.15 82.77 20.69 41.39 9.74 166.67 1591.55 1622.98
0.2 110.36 27.59 55.18 12.98 166.67 1591.55 2163.97
0.25 137.95 34.49 68.98 16.23 166.67 1591.55 2704.96
0.3 165.54 41.39 82.77 19.48 166.67 1591.55 3245.96
0.35 193.13 48.28 96.57 22.72 166.67 1591.55 3786.95
0.4 220.73 55.18 110.36 25.97 166.67 1591.55 4327.94
0.45 248.32 62.08 124.16 29.21 166.67 1591.55 4868.93
0.5 275.91 68.98 137.95 32.46 166.67 1591.55 5409.93
0.55 303.50 75.87 151.75 35.71 166.67 1591.55 5950.92
0.6 331.09 82.77 165.54 38.95 166.67 1591.55 6491.91
As stated in the From F7-F16 Max Driving Force = 0.5 x As motor will be driving two Applying the gear ratio, With the gear With the gear ratio G,
Technical Max Axle Torque/Wheel axles, the required torque G, and the transmission ratio G, the the motor's required
Specification Radius. Rearranged to find output from the transmission efficiency, gives the motor's required speed to acheive the
the max axle torque. This is double the single axle value. required motor output speed to acheive 15km/h target is
assumes torque is split torque. the 15km/h above in rpm, giving
evenly between the two target is above in the average value we
Figure 11: Excel spreadsheet showing performance calculations for specification of traction motors
31
32
Traction Motor and Transmission Requirements Calculations Sheet 2: 3 Hours at 5km/h on a 5% Gradient with 400kg Load Calculations
Figure 12: Excel spreadsheet showing performance calculations for specification of generator
CHAPTER 4. TRACTION
4.5. ELECTRIC DC MOTOR SELECTION 33
(a) Lynch brushed PM type [18] (b) Etek brushless PM type [19] (c) Perm brushless PM type [20]
brushed permanent magnet motors available from Lynch Motor Company (LMC) and Agni Motors
would be most suitable due to their compact ’pancake’ design, straightforward control and high
torque density. The motors are also capable of regenerative braking without any modification, and
the control systems to enable them to do this are relatively simple and commonplace. As the motor
designs from both companies were very similar, the more local company, LMC, was chosen to
supply the motors.
120 Nm.
was readily available. The speed reduction of 5:1 between the sprockets was directly achieved by
having the ratio of teeth to be equal to this value. After consideration of the maximum size of the
larger axle sprocket (which had to be smaller than the wheel diameter), a 50 tooth larger sprocket
was chosen, meaning that a smaller 10 tooth variant was required for the motor shaft. To drive
both axles from a single shaft, there would have to be two of the smaller sprockets mounted on it.
A rendered CAD image of the design, with the motor, chains and sprockets included, can be seen
in Figure 16. The design of keyways on the sprockets was required to ensure correct location and
power transmission. Also shown in the image is a motor mounting which is discussed further in
Section 4.9. Allowance for chain stretching was made by having the mount vertically adjustable so
that the distance between axle and motor shaft centres could be adjusted to take up the slack in
the chain.
(a) Utility Free Living DC3000S [23] (b) Alphagen DCX3000 [24]
they are often smaller, lighter and more quiet than conventional generators, though more expensive
due to their added complexity.
It was decided within the team that the extra expense would be worthwhile for the reduced
noise output and mass, therefore an ‘inverter’ type generator producing 230 V AC was settled
on. A 2.8 kW model was found which is detailed further in Section 4.7.4. To convert the 230 V
AC output into the 48 V DC required for the traction system, a number of options were again
considered. These included creating a rectifier circuit, using a linear DC power supply or using a
DC switched mode power supply (SMPS). The extra steps required to design and manufacture a
custom rectifier circuit, with adequate safety, meant that it was discarded. The comparison between
linear and switched mode power supplies had the SMPS come out far ahead; they are considerably
more efficient, lighter and smaller than equivalent linear power supplies. Therefore a 3 kW single
output SMPS was found, the details of which are discussed in Section 4.7.4. It was chosen as it
was capable of taking an AC input between 180 V and 264 V and outputting at 48 V DC, with a
rated current of 62.5 A.
Figure 18: Generator and SMPS chosen to power locomotive traction system
4.8.1 Requirements
With the motors decided on, a number of different methods of control for permanent magnet DC
motors were considered. The main features needed were variable speed control, the ability to
perform regenerative braking, forward and reverse modes, and ideally a switchable free-wheeling
mode for the energy recovery challenge. It would need to be able to cope with a high current draw
for the acceleration events, likely between 200 A and 400 A.
Due to the close integration with the rest of the electronics and control of the locomotive, these
requirements were passed on to the sub-team working on that area. Detail on the 4QD-300 con-
troller that was chosen for the system can be seen in Section 7.5. This also includes detail of all of
the bench top testing that was undertaken for the traction system before it was assembled into the
locomotive. Further testing, with the traction system driving a load, can be found in Section 13.2.
Figure 21: Motor with machined spacers, small sprockets and mount
The process of developing the entire traction system generated many possible design choices,
some of which were not feasible for this project in its first year. Based on experience gained this
year, areas for further work are outlined in Section 14.
5. Energy Recovery
5.1 Introduction PN
In conventional vehicles, energy is lost through braking. Braking friction extracts the kinetic energy
from the rotating wheels causing the vehicle to decelerate, generating heat and sound [27] [28].
Energy recovery systems aim to reduce this loss.
A Kinetic Energy Recovery System (KERS) can be implemented as an attempt to harvest some
of the kinetic energy lost during deceleration. The energy is then stored in a manageable form from
which useful energy can once again be extracted. This particular style of system is implemented in
Formula One racing cars [29]. The driver is able to store energy extracted from corner braking and
choose to release at key moments for overtaking manoeuvres. However, KERS is also likely to have
strong success in an urban environment where inconsistent traffic speeds are common [30] [31].
There are many considerations with KERS, these lie not only with how to extract the energy from
the moving vehicle but also in choosing how to store the energy. Every time an energy changes
form there are inherent losses [32]. Batteries are commonly used to store energy in automobiles
[33]; however, despite having a large energy density, batteries are heavy in comparison to more
efficient methods. The mass of the KERS should be taken into account when designing the system.
For maximum effectiveness, it is important to ensure that the KERS storage device is able to
store all energy present before braking. For the energy recovery challenge the locomotive will be
coupled to a 600 kg trailing load and so the mass of the train will be a combination of the two.
This total recoverable kinetic energy was estimated at 9115 J. The rail application for the energy
recovery system requires a large power density over that of a large energy density. This factor was
taken into account when researching options for the locomotive as specific power relates to the
acceleration of the vehicle and specific energy is related to the range [34]. The maximum speed
stipulated in the IMechE technical specification is 15 km/h (4.17 m/s) [15]. In the following sections,
the potential KERS systems are investigated and evaluated.
41
42 CHAPTER 5. ENERGY RECOVERY
and flywheel would dictate the maximum rotational acceleration possible. By the same notion, this
gives the flywheel the ability to deliver a large amount of power for a short period of time. Safety is
a serious consideration where flywheels are concerned as a large mass spinning at high speeds
can cause harm if it were to become unstable. A strong housing is required when designing such a
mechanism. Purely mechanical flywheel systems also have the complication of integrating another
mechanical clutch and geared drive system.
Electro-mechanical
The Gyrobus was an electric vehicle used in Switzerland in the mid twentieth century. It utilised an
electrical flywheel in lieu of battery packs. Electromechanical flywheels use electricity to spin up a
flywheel, the reverse operation then enables the rotational energy to be converted back to electrical
energy. This idea was good in principle but the Gyrobus proved to be problematic as minor faults in
the bearings would cause a rapid deceleration of the flywheel resulting in more frequent stops for
recharging of the bus. Not only was this inconvenient but the buses were also far more expensive
than regular fossil fuel vehicles so the project was dropped [38].
Despite this early setback, research and development into more affordable and reliable sys-
tems for commercial use has been taking place [39]. Much larger electromechanical flywheels are
planned to be implemented into the National Grid as a method of storing excess energy, with a
view to discharging the system to increase the Grid output at times of peak demand [40].
Flywheels appeared to be promising as they have the possibility of a larger efficiency than
batteries [41]. The losses in the electromechanical flywheel system occur through contact friction
in the bearings and air resistance opposing the angular velocity. For improved efficiency, the system
preferably requires a vacuum to limit the effects of air resistance during operation. The losses can
further be reduced by implementing magnetic bearings between the flywheel and shaft. This type
of bearing supports its load via magnetic levitation without the need for physical contact, so friction
is eliminated and the mechanical wear substantially reduced [42]. Magnetic bearings are also able
to operate within a vacuum where conventional bearings fail due to lack of lubrication.
For the current project, companies including Torotrak plc and Williams Advanced Engineering
Ltd were contacted with regard to supplying an electromechanical flywheel. However, it proved
difficult finding a supplier able to provide a product suitable for the project requirements and the
quotes obtained were far in excess of the budget. Despite suggested commercial research into
smaller and more compact designs, it was deemed unfeasible to purchase a compact system on
the current budget.
decelerate quickly and accelerate slowly. This reduces energy losses and the possibility of wheel
spin. The options of storing energy in both a rubber elastomer or in a coiled steel spring have been
considered.
Elastic
The energy density of rubber is 11 kJ/kg, 27.5 times greater than that of spring steel [43]. Still elas-
tic (rubber) has its limitations. The low Young’s modulus results in a large material strain needed to
store sufficient energy, approximately 600% strain [43]. Implementing this system requires clamp-
ing the ends of the material, this often includes stress raises and can initiate crack propagation,
leading to fatigue. The elasticity of the material is highly temperature dependant, so different strains
are required in different environments, this poses control integration issues. Finally, the life cycle of
the cable is very unpredictable, even with strain levels remaining within their elastic limit, if the ca-
bles are exposed to an abrasive species such as grit or exposed to UV radiation, then the material
can deteriorate in tens of cycles as opposed to thousands as predicted. To add to these problems,
the loading process is inherently inefficient; during the load/unload cycle a significant amount of
energy is dissipated as heat, as shown in Figure 22.
Spring
Spring energy storage is similar to a bungee cord in that energy is stored as potential strain en-
ergy, though the limitations are very different. Steel can function in a far wider temperature range,
it possesses a superior, more predictable cycle life which makes it a far safer material to work
with. The principal issue with spring systems is that the energy density of steel is particularly low,
0.4 kJ/kg [43]. For the application specified in Section 5.1, 21 kg of steel wound into coils or springs
would be required to provide 9115 J of energy storage.
store the compressed fluid for use in energy storage and recovery [46].
The problems foreseen with this type of system were mainly thermodynamic in nature. It was
predicted that the chosen fluid would increase in temperature under compression and so the system
would need suitable insulation, otherwise this thermal energy could be emitted and lost to the
surroundings [47]. Also, the sound emitted from the pump is wasted energy and may be problematic
in the noise control assessment. An additional factor included the security of the system as a small
leak at any point would be difficult to detect and could cause significant pressure loss. As with
elastic systems, controlling the release of the pressurised system would also present problems.
Although a high pressure to begin with would be beneficial, it would quickly drop and become very
ineffective. Being difficult to achieve a perfect system, the notion of hydraulic recovery was dropped
for this design, however this KERS could be trialled by others in the future as a potential KERS.
Figure 23: Ragone plot displaying the specific power and specific energy various storage mediums
[53]
In order to evaluate the suitability of each energy storage method for the task required, a number
of important considerations are explored:
• Cost • Safety
• Cycle Life • Scalability
• Drive Integration • Specific Power
• Control Integration • Specific Energy
• Efficiency • Stored Energy Retention
• Reliability and Simplicity • Weight
To quantify the relative importance of each of the design considerations listed above analysis
carried out in the form of a binary matrix is displayed in Table 5. Stored Energy Retention
Normalised Score
Drive Integration
Specific Energy
Specific Power
Product Score
Scalability
Reliability
Simplicity
Efficiency
Cycle life
Weight
Safety
Cost
Table 5: Binary weighting matrix showing relative importance of energy storage medium parame-
ters
Once the relative weighted importance of each design consideration was found, analysis of
each storage method was carried out. Each method is ranked with regards to the others in how
well it meets the specific design criteria. The weighted importance of each consideration, defined
in Table 5, was then applied and the final energy storage method score evaluated, the results of
which are shown in Table 6.
5.3. SUPERCAPACITORS 47
Drive Integration
Specific Energy
Specific Power
Overall Score
Scalability
Reliability
Simplicity
Efficiency
Cycle life
Weight
Safety
Cost
Battery 7 4 3 7 3 7 7 6 7 7 5 7 3 5.66
Elastic 1 7 1 1 1 1 1 1 5 2 2 1 7 1.87
Flywheel (electrical) 5 1 5 5 4 4 5 5 4 5 3 5 4 4.27
Flywheel (Mechanical) 4 5 6 4 6 5 3 4 3 4 4 6 5 4.53
Hydraulic 3 2 4 3 5 2 4 3 1 3 6 3 1 3.46
Spring 2 6 2 2 2 3 2 2 2 1 1 2 2 2.19
Supercapacitor 6 3 7 6 7 6 6 7 6 6 7 4 6 6.02
It can be seen from inspection of Table 6 that the method with the overall highest score is the
supercapacitor and is therefore the chosen energy recovery storage method.
−t
VC (t) = VC (0) + (Vin − VC (0))(1 − e RC ) (5.3.1)
where VC (0) is capacitor voltage at t = 0s, Vin is voltage supplied by source, R is circuit series
resistance, C is rated capacitance of capacitor and t is time. Rearranging Equation 5.3.1 we achieve
a simplified equation of capacitor voltage with time, given below [57]:
−t
VC (t) = Vin − (Vin − VC (0))e RC (5.3.2)
Using the I = V /R relation to give the current through the circuit as a function of time IC (t) [59]
[57]:
Vin − VC (0) −t
IC (t) = e RC (5.3.3)
Z
48 CHAPTER 5. ENERGY RECOVERY
Using this, the energy dissipated as heat in the resistor as a function of time EVin → Z (t) can
be given by integrating the value of power of the resistor [57], found using P = VI :
t
C(V C (0) − Vin )2
Z
−2t
EVin → R(t) = VR IC (t)dt = (1 − e RC ) (5.3.4)
0 2
and the total energy being either dissipated in the system, or stored into the capacitor EVin → RC (t)
is given by using the P = VI integral as above [57]:
Z t
−t
EVin → RC(t) = Vin IC (t)dt = CVin (V C (0) + Vin )(1 − e RC ) (5.3.5)
0
The energy being stored in the capacitor as a function of time EVin C(t) is given by [57]:
−t
EVin → C(t) VC (0) + Vin + (VC (0) − Vin )e RC
ηC = = (5.3.7)
EVin → RC(t) 2Vin
Upon inspection of this equation it is apparent that to increase the efficiency of charge of the
capacitor you can either increase the initial charge voltage when t = 0 or increase the ratio of
VC (0) : Vi n
Efficiency of the charging cycle also increases as VC (t) approaches Vin . This is due to the
proportion of energy stored in the capacitor compared to energy dissipated in the resistor increasing
as time progresses. Being at a maximum at a point in time where VC (t) = Vin ; t = ∞.
To achieve ideal charging efficiency:
VC (0)
→ 1; t → ∞ (5.3.8)
Vin
If at the beginning of charge VC (0) = Vin then the efficiency of the charge cycle will be 100%.
This is because the initial current will be zero, and therefore no energy will be dissipated through the
resistor. This situation, although ideal in efficiency terms, places limitations upon energy storage
ability, as the energy stored in a capacitor between two voltages VC (0) and VC (∞) is given by
Figure 5.3.9, adapted from [37]:
C(V in − VC (0))2
EVin → C = (5.3.9)
2
where VC (∞) = Vin .
This means that if VC (0) = Vin then Vin − VC (0) = 0, so the energy stored in the capacitor is zero,
therefore useless for the intended purpose. It is apparent that a compromise must be met at the
cost of efficiency, so that the required energy storage capabilities can be met. In order to achieve
5.3. SUPERCAPACITORS 49
• In order for the efficiency of charging the capacitor to be maximised within the experimental
requirements, the value of VC (0) : Vin can be increased by increasing the capacitance, and
therefore the energy capacity.
• The optimum pre-charge of the capacitor is the maximum value where the required energy
capacity is still available for Erecov . The maximum value of Vin will depend upon safety mar-
gins, but will be maximised within that to ensure efficient energy storage.
• The value of VC will be optimised to ensure maximum efficiency, whilst still maintaining an
acceptable charge rate.
Figure 24 shows the theoretical charging efficiencies of the supercapacitors when ‘fully charged’
over a range of Vin values when Vfull is equal to Vin . The supercapacitors are defined as ‘fully
charged’ after five time constants; when t = 5(RC). At this point in time the value of VC will be
within 1% of Vin , and therefore can be treated as full.
50 CHAPTER 5. ENERGY RECOVERY
Figure 24: Charge efficiencies of supercapacitors of different capacitance values with variable Vin
values
Upon inspection of Figure 24 it is apparent that increasing the capacitance value of the system
results in increased charge efficiency. The magnitude of this efficiency increase from 19.3 F to 83 F
is large in comparison to the change from 83 F to 165 F. In this particular case, where Vfull = Vin , it
is beneficial not to use the 19.3 F system as the efficiency disparity is large, whereas the benefits
of using the 165 F module over the 83 F module are small enough to warrant further investigation.
It is worth noting that the plots shown in Figure 24 show the efficiency of charge of the individual
capacitors at the point in time where the capacitor is defined as full. These plots display ηC at 5RC,
and are only true for that particular point in time, for that particular Vin value. Due to the fact that the
charge efficiency improves over the duration of a charge cycle, the total energy transfer efficiency
into the capacitor will not be as high as shown, and therefore the efficiencies are overestimated.
That being said, the disparity of charge efficiencies from t = 0 to t = 5RC are 0.06$, 0.15%, and
0.48% for the 48 V 165 F module, 48 V 83 F module and the three 16 V 58 F modules respectively
for Vin = 48 V , and therefore is not considered to be significant.
Below, the effect of varying the ratio of Vfull : Vin upon the charging efficiency is investigated.
This is achieved by setting Vfull at values of Vin − [0.5, 1.0, ..., 2.5, 3.0]V Altering the value where
the supercapacitor is considered full results in a corresponding change in VC (0). This variation of
pre-charged state is was investigated for different values of Vin for each of the three supercapacitor
arrangements investigated; 19.33 F, 83 F, and 165 F, which is shown in Figure 25.
5.4. ENERGY RECOVERY CHALLENGE: INITIAL MODEL 51
Figure 25: Charge efficiency of 48 V 165 F module with variable Vfull and Vin values
Similar graphs were also obtained for the other two modules and analysed.
It is apparent that with an increasing value of Vfull there is a corresponding increase in charging
efficiency. Within the likely operating range of the system the charge efficiency changes in a linear
fashion with both an increase in the value of Vfull and Vin . Importantly the change in charging
efficiency has a fairly uniform increase of around 1.45% when comparing the 165 F module to
the 83 F module with a comparable Vfull and Vin values. This may not appear to be of large
consequence but due to the importance of efficiency of energy recovery even a small increase in
charging efficiency is highly desired.
From this comparison of the three different supercapacitor modules investigated, it is apparent
that the 48 V 165 F module is the optimum choice for efficiency of charge purposes.
Figure 25 illustrates the effect of varying Vfull and Vin upon charging efficiency. It can be seen
that increasing the value of Vfull from 3 V below Vin to Vin yields an efficiency increase of 5.8%
compared to the increase of 16.3% when increasing the value of Vin from 40 V to 48 V. This
highlights the importance of selecting the ‘correct’ Vfull and Vin values when considering the safety
limits of using high values of Vin and when considering the value of Vfull : Vin which dictates the rate
at which the supercapacitor can be charged.
The design of the supercapacitor energy recovery circuit will be covered later in this report in
Section 5.5.
Assumption Value
Supercapacitor Efficiency 0.9
DC Converter Efficiency 0.85
Transmission Efficiency 0.9
Motor Efficiency 0.9
Overall Efficiency 0.43
Mass of system 1050 kg
Transmission Ratio 4.75
Maximum Acceleration 1.3 ms−2
Maximum System current 400 A
Wheel Diameter 0.25 m
Max Velocity 15 kmh−1
Figure 26 models the velocity of the train during the energy recovery challenge; the locomotive
starts at 4.16 m/s, decelerates at 1.3 m/s2 until stationary, can be modelled to remain stationary for
0.5 seconds, and then accelerates using the recovered energy.
The locomotive has been modelled to deceleration rate at 1.3 m/s2 , this is the maximum allow-
able deceleration rate set by the IMechE railway challenge authorities. Using the equation F = ma,
and v = u + at, the velocity time graph for the locomotive are plotted (see Figure 26).
The acceleration of the train through discharging the supercapacitor can be calculated using
the minimum of the two term in Equation 5.4.1.
µR P
a = min , (5.4.1)
m mV
Where µ is the coefficient of rolling friction, P is the supercapacitor power output, m is the
combined locomotive and trailing mass, R locomotive reaction force and V is locomotive velocity.
The first term calculates the traction available for the wheels on the rail, and second term calculates
the maximum available power output of the supercapacitor. The maximum output of the capacitor
is limited by the overall resistance within the circuits or Equivalent Series Resistance (ESR). It is
initially estimated at 250 mΩ. This limits the maximum electrical power through the system to be in
the region of 8 kW.
Traction is independent of speed, as the rolling friction coefficient ref µ is velocity independent.
This gives the maximum acceleration of the train to be 1.45 m/s2 . Traction dominates behaviour at
low speeds, at higher speeds acceleration is limited by power. Equation 5.4.1 calculates these two
acceleration rates and returns the minimum.
5.4. ENERGY RECOVERY CHALLENGE: INITIAL MODEL 53
Figure 26: Velocity time graph of the locomotive during the energy recovery challenge. Model has
been calculated considering inefficiencies of each component as stated in Table 7
.
Figure 26 shows that the supercapacitor can deliver sufficient power such that the locomotives
acceleration is dominated by friction. This highlights that there is a real possibility of wasting power
through wheel spin, which is an important design consideration for the motor controller.
This model also simulates the current and voltage which the locomotive will produce during
regenerative braking and required during acceleration.
The motors accelerate and decelerate the train with a constant torque and therefore constant
current, 105 A and 84 A respectively. The supercapacitor voltage is modelled such that the mod-
ule is pre-charged to 44 V, this is done to increase the efficiency of the system, as explained in
Section 5.3.1. We can see in Figure 27b, the supercapacitor voltage rises from 44.0 V - 45.04 V.
This small increase in voltage is desired so that the supercapacitor’s energy input is always in the
efficient energy storage voltage range.
This model has given us the voltage and current produced by the motors during regenerative
braking. It has also calculated some dynamic mechanics, the acceleration rates, and forces re-
quired by the motors. A more detailed and refined simulation of the system is produced in Simulink,
see Section 5.7. The electronic systems in the locomotive design commences using the parame-
ters generated from this initial model.
54 CHAPTER 5. ENERGY RECOVERY
5.5.1 Introduction
The electrical power distribution circuit will control the power flow from the generator and superca-
pacitor to the motors. The system incorporates a generator to produce the DC electrical energy, a
supercapacitor and a 4QD motor controller to control the power delivery to the motors. The super-
capacitor will firstly recover energy when braking and be used to deliver an extra burst of power to
the motors. For effective power flow from the generator to the rails an efficient electricity distribution
circuit is required. The electrical power circuit or 48 V circuit is required to be safe, controllable and
most importantly available to deliver power on demand.
The generator can provide 3 kW of power, when the train is laden under a 600 kg the superca-
pacitor is needed to maintain a constant speed of 15 km/h while climbing a 5% incline, and when
accelerating. These two situations are displayed in bold font in Table 8.
Earthing
Earthing the system components such as the frame, the supercapacitor case, and the shell stops
the build up of static and to prevent components becoming live in the event of a live wire touching
the frame. These components need to be earthed to zero potential.
One earth point is needed from the 48 V circuit to the frame, following this an earth strap is
needed from the frame to the bogies. If there were several different earth points on the frame,
small currents will flow between the earth points, the frame would act as an antenna and will
5.5. ELECTRICAL POWER DISTRIBUTION 55
interfere with the electronics of the myRIO and motor controller. Thus it is important to earth all the
components at the same point on the frame, the current is taken to the earth point via an insulated
wire not through the frame.
Initially it was intended to connect the bogies to the axle via graphite brushes. This would
ensure a good connection to the axles. A simple test with a multimeter proved the axle bearings to
be conductive, concluding graphite brushes are not needed and a earth strap from the frame to the
bogies will be sufficient.
Supercapacitor Safety
The supercapacitor has the ability to store large amounts of energy (190 kJ); this energy needs to
be safely monitored and controlled. The mechanisms of safely controlling the capacitor’s charge/dis-
charge cycle will now be discussed.
The supercapacitor cannot be charged over a certain voltage, the myRIO constantly monitors
the supercapacitor voltage and is programmed to stop charging the supercapacitor before the volt-
age is at full capacity, 47 V.
In the case of connecting the generator in reverse polarity, a bypass connector circuit is added
to the circuit. This allows current to bypass the capacitor and blow a fuse, preventing the superca-
pacitor from being damaged, this circuit adaptation can be seen in Figure 28.
The supercapacitor’s maximum capacity is 48 V; however it will not be charged to more than
44 V by the generator to leave some capacity for the recovered energy to be stored. The energy
needed to overcharge the capacitor from 44 V is calculated in Equation 5.5.1.
1 2 1
C[Vfull − V 2 ] = 165[482 − 442 ] = 60.7 kJ (5.5.1)
2 2
To put that into context, the train’s kinetic energy at full speed is 7.8 kJ. This is only 12.8%
of the energy needed to take the supercapacitor into a state of overcharge. The supercapacitor
maybe be overcharged if the train is rolling down a long steady slope. The recovered energy of
the train rolling down a hill of height h is calculated by E = mghη, where ηERS is the efficiency of
the energy recovery system 43%, as calculated from the inefficiencies in Table 7. The maximum
descent which the train can safely regenerate down is calculated in Equation 5.5.2.
E 60700
h= = = 16.0 m (5.5.2)
mgηERS 900 × 9.81 × .43
Equation 5.5.2 does not take into account the friction and air resistance losses which will inher-
ently make the ERS more inefficient. The locomotive will be tested at Eastleigh Lakeside Railway
and Stapleford Miniature Railway in Leicestershire, a decent of this magnitude does not exist at
either of these locations. Therefore, neither does the opportunity to overcharge the supercapacitor
in this manner.
Fail-Safe Design
The train needs so be safe in the event of control loss. This requires control over the supercapacitor
and being able to manually stop the train. In the event of power loss the brakes will automatically
come on. The generator and supercapacitor will still have the ability to supply power. Accordingly,
the system will be fitted with solenoid switches which return the system to a certain state when the
power to the control circuit is lost. This will stop the power to the train and discharge the capacitor.
56 CHAPTER 5. ENERGY RECOVERY
There is also a manual switch to serve the same purpose, which the operator will have access to.
The supercapacitor is fitted with a discharge circuit, as seen in Figure 28. This is simply a
large resistor that is capable of dissipating the potential energy safely as heat. When power is
lost to the switches, the dissipation resistors are automatically connected and the supercapacitor
is discharged. This will leave the circuit dead and safe to work on.
The dissipation resistors themselves are a safety hazard, they heat up to high temperature
when the supercapacitor is discharging. Although the resistors dissipate heat quickly, they will rise
to a considerable temperature. As explained in Section 5.6.4, the resistors require air flow over
them and to be positioned so that operators cannot touch them.
V 48
I= = = 7620 A (5.5.3)
R 0.0063
A charge current of 7620 A would cause severe sparking and damage to the generator. Two
options are considered to reduce the charge current to below 62.5 A: variable voltage charging and
soft charging.
The current of 7620 A, is known as the short circuit current; this is assuming no circuit resis-
tance. In reality there will be a significant amount of circuit resistance, and as the supercapacitor
charge increases the voltage drop between the generator and capacitor reduces. These two factors
lower the current flow through the capacitor.
The variable voltage charging method involves increasing the supply voltage to the supercapacitor
as it charges. If the voltage drop between the SMPS and supercapacitor is below a calculated
value, the current flowing will be within the generator’s rated voltage. This voltage drop is calculated
in Equation 5.5.4.
If the voltage difference between the two components is less than 0.39 V, the SMPS will be op-
erating within it’s rated current. Note this calculation neglects the ESR of other circuit components,
giving an underestimate for the maximum voltage drop.
This setup is very efficient as almost all of the energy supplied is stored in the capacitor. The
concern for this method of charging is that there is little margin for error. The supply voltage needs
to be carefully monitored as the supercapacitor voltage increases. If the voltage drop closes to less
than 0.39 V the charge time will be increased. If the voltage drop exceeds 0.39 V, the SMPS will
experience a current above what it is rated for. This is damaging to the SMPS and dangerous for
users [60].
5.5. ELECTRICAL POWER DISTRIBUTION 57
The alternative option to charge the supercapacitor is to set the SMPS to full power, and use a
resistor to limit the current around the circuit. The resistor is put in series between the SMPS
and the capacitor to limit the current. Again using Ohm’s law it can be found that the minimum
resistance required to limit the 48 V SMPS’s current to 62.5 A, is 0.77 Ω.
A 1 Ω resistor can be used to limit the current to 48 A when the generator is at maximum power.
This is a safe option as there is no possibility of a charge current above 62.5 A.
There are a few issues using this method of charging; firstly the capacitor takes longer to charge,
and secondly it is a much less efficient way of charging the capacitor.
As the current is limited from 62.5 A to 48 A, it takes longer for the capacitor to charge. In
addition, when the capacitor is nearly full, such as 40 V while being charged with a 48 V supply
voltage, there is only an 8 V drop between the two components. This limits the current to just 8 A.
In this instance the generator is only delivering 64 W of power, utilising very little of the SMPS’s
potential 3 kW of power.
Additionally a large amount of energy is wasted as heat, as P = I 2 R where R = 1 Ω and initially
I = 48 A. At the start of the charging process 2.3 kW of heat energy is dissipated. This is reduced
as the capacitor’s voltage increases.
Although this method of charging is slower than variable voltage charging, it charges the capac-
itor in an acceptable length of time. The charge time is equal to 5 time constants:
Figure 28: Circuit diagram of the main power system within the train
The circuit displayed in Figure 28 is designed such that power can be drawn from the superca-
58 CHAPTER 5. ENERGY RECOVERY
pacitor and the generator without any changes in the control circuit. The start up and shut down
procedures and the four different operating scenarios of the locomotive will now be discussed,
stationary, regenerating braking, fast acceleration and cruising.
1. Start up: The supercapacitor has no charge stored. The supercapacitor will need to be
charged through the soft charging resistor. It is recommended to charge the capacitor to 90%
of its capacity, 44 V. This is achieved by setting the DC SMPS to 44 V.
2. Capacitor Charged: When the supercapacitor is charged to 42.5 V, within 1.5 V of the supply
voltage, disconnect the soft charge switch. The soft charge resistor can be by-passed to
reduce charging time.
3. Operational: The train can stay in this state and run at any speed controlled by the motor
controller. The four operating states will now be explained.
(a) Stationary: The motor controller does not draw any power from either the generator or
the supercapacitor.
(b) Cruising: The train is drawing less than 3 kW of power. The generator is producing a
sustainable amount of power. The supercapacitor charge state will stay constant.
(c) Accelerating or Climbing: The locomotive is consuming >3 kW of power. The gener-
ator cannot supply more than 62.5 A, the generator voltage will start to dip below 44 V,
the supercapacitor charged at 44 V will supply the extra required current to the mo-
tor controller. When the climb is finished, the supercapacitor will be charged with less
than 44 V. The generator will be recharging the supercapacitor back up to 44 V while
simultaneously powering the train.
(d) Regenerating: The locomotive is decelerating and the motor controller is not drawing
any current. The motors are producing a current back through the circuit in the oppo-
site direction. This current is directed through the motor controller into the top of the
capacitor. The current cannot go into the generator as it is protected by the diodes.
The supercapacitor stores this charge, making the charge state of the supercapacitor
approximately 45 V. When the locomotive starts moving forward again it draws current
from the place of the highest voltage, i.e the supercapacitor. The locomotive will run off
regenerated energy until the capacitors charge state drops to 44 V when it will naturally
switch to using the generator’s power.
4. Shut Down: Disconnect the overcharge switch, to disconnect the supercapacitor from the
power circuit. Connect the discharge resistor to the supercapacitor, automated discharge
switch, the current will flow through the resistor and dissipate the energy away as heat.
5. Emergency Stop: Push the emergency stop switch, not displayed in Figure 28, as it is built
into the 12 V control circuit. This switch is mounted on the outside of the train. When pressed,
power to the motors is stopped, the supercapacitor discharges, and the brakes are activated.
5.6. COMPONENT SELECTION 59
5.6.1 Switches
Switches control the operating states and start up and shut down procedures of the locomotive.
The switches need to compatible with the myRIO, which can send a logic output signal from 0 V to
5 V. This signal is used to control the switches. The function and rating of each of the switches are
described in Table 9.
Of the switch manufacturers considered, the Albright SW80 series switches have been selected
in preference to the Amtech Ltd switches. The Albright SW80 switches are better suited to this
purpose as they are simple and readily available. They can be powered by the 12 V batteries and
controlled by the myRIO.
Q = mc∆T . (5.6.1)
The resistor material is taken to be a ceramic and the specific heat capacity, c, is estimated to
be in the lower region of ceramics (approximately 0.7 kJ/kgK) [61]. The mass of the resistor is
represented by m and taken to be 1 kg, then the change in temperature is ∆T. It was assumed that
the starting temperature was an ambient 298 K (25◦ C) and so the final temperature of the resistor
was calculated to be 374 K (101◦ C).
The electronic power supplied by the supercapacitor is to be dissipated by the resistor as heat.
It is assumed that the resistor is heated uniformly and the air temperature at the surface of the
resistor is taken to be equal to that of the resistor. Newton’s law of cooling is given as [47]:
Q̇ = As h∆T . (5.6.2)
Where As is the surface area of the resistor, h is the convective heat transfer coefficient and ∆T
is the change in temperature between the air at the surface and the surrounding air (Ts -T∞ ). It
is reasonable in assuming the resistor to be a perfect a cylinder. From this, the heat transfer
coefficient can be estimated from determining the Nusselt number, Nu. The Nusselt number is
dependent on the orientation of the resistor and for this project will be mounted horizontally [47]:
( 1/6
)2
0.387RaD
Nu = 0.6 + . (5.6.3)
[1 + (0.559/Pr )9/16 ]8/27
5.6. COMPONENT SELECTION 61
The above equation expresses Nu for a horizontal cylinder, where Ra is the Rayleigh number
and Pr is the Prandtl number. The Rayleigh and Prandtl numbers are calculated to be 0.36 and
1.3×10−5 respectively. The Nusselt number for this scenario is therefore estimated to be 0.43.
The relationship between the Nusselt number and the convective heat transfer coefficient is given
as [47]:
hLc
Nu = . (5.6.4)
k
Lc is the characteristic length which in this case is the diameter of a horizontal cylinder and k is
the thermal conductivity of the air (taken to be 1.4) [37]. Substituting Equation 5.6.4 into Equation
5.6.2, Newton’s law of cooling becomes:
kAs Nu
Q̇ = (Ts − T∞ ). (5.6.5)
Lc
The surrounding air temperature, T∞ , is assumed to be the ambient temperature of 298 K. Ts is the
surface temperature calculated from equation 5.6.1 as 374 K. The cooling power of the surrounding
air is determined to be 86.6 W, which is multiplied by the sample discharge time of 60 s to ascertain
the energy transferred into the surrounding airflow (5.2 kJ). Using equation 5.6.1, but accounting
for the energy lost through cooling, the new maximum temperature of the resistor is found to be
approximately 94.0◦ C.
This temperature is acceptable and will not interfere with surrounding systems. It should be
noted that these previous calculations only represent natural convection. Vents will be cut into the
shell to allow a mass transport of air through the locomotive incurring further forced convection
during travel and the cooling fins on the actual resistor will increase the surface area. Allowing the
supercapacitor to discharge through a resistor and a fan in future would also help to create a forced
convection.
– The cap protecting the terminal from accidental contact is securely fastened onto the
supercapacitor by a bolt tightened beyond finger tightness, therefore requiring tools to
gain access. The intention is to reduce impulse, ill-informed access.
• Accidental contact with exposed connections/cabling must be made to be as difficult as pos-
sible.
– When in place on supercapacitor, the only access to the terminal is the cable inlet/outlet.
62 CHAPTER 5. ENERGY RECOVERY
– Opening dimensions are constrained to reduce clearances on either side of the cabling
as much as possible, see Figure 29a.
– Standardisation of the cable approach angle to the terminal allowed cabling clearances
to be further reduced from initial designs.
• Cap structural integrity should not be compromised by realistic eventualities such as the im-
pact of a falling tool.
– Acrylonitrile Butadiene Styrene (ABS) over the alternative Polylactic Acid (PLA) due to
its higher impact resistance: 3.04 J/cm compared to 0.961 J/cm [63]
– Stress raising features of the design are limited as much as possible, with wall thickness
maximised (minimum 4mm), and sharp corners avoided. Also, a high fill density was
implemented, the physical difference of which can be seen in Figure 29b, to reduce
internal stresses and to increase the strength of the cover.
• Cap material must be non conductive, and also resistant to expected ambient temperatures.
Shown below in Figure 29a is the final design of the supercapacitor protective cap, and in
Figure 29c the caps can be seen on the terminals.
5.7 Simulation PN
5.7.1 Aim
Simulink R2014a is a programming language developed by MathWorks which is used for simulating
models of dynamic systems through the use of block diagrams [65]. The primary aim for using this
modelling software was to ensure that the proposed electrical circuit designs would be consistent
in operation. Before the construction of the locomotive was fully completed, this model showed
what can be expected to occur during operation.
Once construction is complete, the secondary aim is to compare the true locomotive operation
against that of the simulation. The simulation is hoped to provide near identical results to the final
tested operation with a view to altering the Simulink circuit prior to any physical design changes.
It is hoped that the process detailed here will provide a platform for future additions. An accurate
model can provide a cheap solution for preliminary testing of any future ventures. Over the following
Section 5.7 the energy recovery simulation design procedure and results will be presented.
DC Motors
DC machines specified with permanent magnets were chosen to represent the proposed Lynch
motors. Preliminary calculations were to assess the forces and torque loads that the motors would
have to overcome. The total force required to accelerate the train is a function of the train mass, m,
and the desired acceleration, a: F = ma. The total torque, T, required to move the train was then
found by multiplying the above result by the radius, r, of the wheels: T = Fr . The constituent parts
of the torque equation were then displayed with respect to the change in angular velocity, ω:
dv dω
T = mr = mr 2 . (5.7.1)
dt dt
Finally, the gearing ratio between motors and axle was taken into account and the coefficient of
angular acceleration equated to the moment of inertia, J:
1 dω dω
T = mr 2 =J . (5.7.2)
G dt dt
The above set of Equations 5.7.1 and 5.7.2 detail the method for identifying the moment of inertia,
mr 2
G , required to overcome to move the mass of the train. However, other rotating components such
as the axles, gears and wheels also have moments of inertia that need to be overcome. Where not
provided by data sheets these inertias, J, are approximated by taking each component to be a solid
cylinder and using J = 0.5mr 2 where m is the mass of the object and r represents the radius. The
individual values were calculated and combined to form a single inertia as seen in Table 10. This
value was then increased to 5 kg/m2 to allow for trailer wheels and axles, then halved and input as
64 CHAPTER 5. ENERGY RECOVERY
It was largely assumed that the locomotive would be travelling on a horizontal surface, however
additional torque loads could be applied for when the locomotive would be faced with a gradient.
The additional force term is determined through: F = mgsinθ, where the mass of the train is given
by m, the acceleration due to gravity is g, and angle of incline given by θ [37]. Multiplying this force
by the wheel radius, the additional torque load can then be found as the result of the train situated
on an incline. For a 5% gradient, the corresponding torque load is found to be in the region of
32.1 Nm per motor. This can be applied as an input to simulate the train ascending up an incline.
The armature resistance and inductance were taken from the motor specifications and the vis-
cous friction coefficient was taken to be zero because no substantially lubricated bearings are in
use. The Coulomb friction torque within the motor was estimated by multiplying the no-load current
(7 A) by the motor constant (0.113 Nm/A). The rolling resistance was calculated by taking a product
of the rolling resistance coefficient of 0.04, determined in Section 4.3.3, and the weight of the train
(mg). This was then multiplied by the wheel radius to give a rolling resistive torque of 5.15 Nm.
After combining this value to that of the motor, the result was then approximately extrapolated to
account for the entire drivetrain and the Coulomb friction torque estimated at 10 Nm. It is hoped that
in operation all mechanical connections will rotate and mesh freely. Finally, to simulate a stationary
start, the initial speed current were set to zero.
Parameter Value
Armature Resistance (mΩ) 24
Inductance (µH) 23
Motor Constant (Nm/A) 0.113
Total Inertia (kgm2 ) 2.5
Viscous Friction Coefficient (Nms) 0
Coulomb Friction Torque (Nm) 10
Initial Speed (rad/s) 0
Initial Field Current (A) 0
Supercapacitor
A supercapacitor module is constructed from an arrangement of cells; their orientation was de-
termined from knowing the individual properties of each cell. For the Maxwell supercapacitor, the
additional Stern-Tafel parameters were left predetermined. The primary parameters required for
the supercapacitor fields can be found in Appendix D, however, the key values are summarised in
the following Table 12.
5.7. SIMULATION 65
Parameter Value
Rated Capacitance (F) 165
Equivalent Series Resistance (Ω) 6.3×10−3
Rated Voltage (V) 48
Surge Voltage (V) 51
Number of Cells in Parallel 1
Number of Cells in Series 18
Leakage Current (A) 5.2×10−3
Initial Voltage (V) 7
Operating Temperature (Celcius) 30
The initial voltage at the start of use will be 0 V and the supercapacitor will have to be soft
charged by the generator before use. The supercapacitor voltage will be set by the operator prior
to the energy recovery task. For the simulation, it is important that the voltage is set below what the
motors are capable of charging. Through numerous Simulink trials it was found that the maximum
initial voltage of the supercapacitor could be 40 V and above this value the motors would immedi-
ately draw from the supercapacitor rather than charge. It is hoped that in true operation the motor
controller should compensate for this and charging at a high starting voltage should be possible.
Motor Controller
The motor controller was represented by means of a simple four quadrant H-bridge with the diode
and switch arrangement governed by a feedback loop. When applying the control feedback, the
desired rotational velocity of the motor shafts was compared to a reference velocity. This angular
velocity reference was taken to be the product of the maximum speed of the locomotive (4.17 m/s)
the gearing ratio and inverse wheel diameter. The comparison of the recorded signal and reference
signal created a velocity error signal. Further feedback in the form of current was also used to
provide the necessary damping. Current is the derivative of the acceleration as determined in the
following equation: T = KM I = J dω
dt , where KM is defined as the motor constant. Therefore, the gain
J
of the angular velocity signal is found to be equivalent to KM = 45. The current signal is integrated
and the combined error signal then passed through a further gain and limiter before entering a
pulse-width modulated (PWM) generator with a four pulse output. The switching frequency of the
PWM generator was made to match the motor controller’s 20 kHz [66]. This final signal is then
supplied to the universal H-bridge. The approximately optimal gain of the signal was established
through multiple tests and 10 gave the system a reasonable response. High gains would cause
the velocity to overshoot and hence the locomotive would be breaking the speed limit set by the
challenge requirements.
Regenerative Braking
For the first 60 seconds of this simulation the motors draw current from the 48 V generator. After
this point the generator is disconnected and the supercapacitor is connected by the use of ideal
switches. There must be a small time lapse between the generator disconnecting and the super-
capacitor connecting to the motors to prevent the generator charging the supercapacitor. A "free-
wheeling" diode was added to the circuit to allow current flow during this time and prevent voltage
spikes when the generator is disconnected from the circuit. For the best results the supercapacitor
voltage is initially set at 7 V. The supercapacitor module then draws a current from the motors and
is allowed to charge. Once at a full state of charge, a voltage monitoring block signals a switch and
the supercapacitor is disconnected. Another block then simultaneously activates the brakes. The
braking aspect of the simulation consists of a torque value (calculated to be 85.3 Nm) multiplied by
a sine function of the velocity output. The end result is that the braking torque will resist movement
in whichever direction the locomotive is travelling. To represent approaching an incline, another
relay block is connected to the distance output. The incline force can be set to occur at a certain
distance from the starting location. The circuit is displayed in Figure 30. A rudimentary method of
controlling current involves the use of variable resistance. Introducing a resistor of 0.06 Ω into the
recovery circuit (see Figure 30) was found to reduce the current range to a more amicable level.
Several hundred amps can be experienced by the controller for a brief period of time as mentioned
in Section 4.8. However, the lower the starting voltage of the supercapacitor, the higher the current
drawn as a result of the back emf of the motors. A large current is initially drawn to create a large
torque to overcome the system inertia, but a smaller amount is expected once the vehicle grains
momentum. This lower current is what is required to overcome the Coulomb friction present in the
drivetrain. The supercapacitor current decreases sharply once the module is connected into the
circuit and quickly returns to zero again. At this point the voltage has been capped as the module
is disconnected from the circuit.
Figure 31: The change in voltage of the supercapacitor with respect to the time taken to regenerate
In Figure 31 the supercapacitor voltage increases as the motors charge the module. After pre-
liminary simulations it was found that the voltage would increase to a certain point and then begin
to decrease as the current flow reverses. To prevent the supercapacitor discharging during braking
a relay and switch are used to disconnect the supercapacitor once it has reached its maximum
state of charge. This state of charge is dependent on the initial voltage set in the parameters. In
reality the motor controller requires a power source for operation and so the supercapacitor will not
be disconnected but the setting changed to allow for “free-wheeling”.
5.7. SIMULATION 67
Figure 30: Circuit diagram detailing the arrangement of components used to simulate the locomo-
tive energy recovery system
68 CHAPTER 5. ENERGY RECOVERY
Figure 32 shows current into the motors to match that of the generator initially and reduce to a
low value as the inertia of the train results in a lower operating current. The current is also seen
to decrease sharply as the supercapacitor is connected. These responses are expected and show
the circuit is working correctly. Finally, Figure 33 shows the velocity of the train. As can be seen,
the velocity increases as the generator supplies the motors and then starts to decrease as a torque
is applied from current supplying the supercapacitor. The train then undergoes braking until at rest.
The sudden secondary increase in velocity once the train stops should not occur and must be the
result of a discontinuity. However, it is ignored as it does not affect the resulting energy stored in
the supercapacitor.
Figure 33: The velocity of the train during acceleration and regenerative braking
This simulation takes the final parameters of the last circuit as initial values. The supercapacitor
voltage is set to the peak value attained in the previous simulation (14.74 V) and a relay block is
connected through a switch. The threshold is set so that the supercapacitor is disconnected once it
reaches the initial charge state of the previous simulation (7 V), meaning the energy utilised by the
motors is only that which was recovered from earlier braking. The free wheeling diode is still present
to allow current flow when the supply is disconnected. As seen from Figure 35, the voltage of the
supercapacitor does not stay constant after disconnection from the circuit. The same behaviour
can be seen to occur during charging (see Figure 31) and it is assumed that this is a result of
internal resistance, RInternal , within the supercapacitor. A realistic representation of the voltage, V,
across the supercapacitor can be given by the equation: V = Vcapacitor + IRInternal . As the current, I,
rapidly approaches zero after disconnection, the measured voltage will either increase or decrease
slightly. As mentioned previously, a discontinuity in the simulation is encountered when the train
velocity reached zero. After coming to a stop the velocity sharply increases and then decreases
and this peak can be seen in Figure 33. This anomaly was difficult to remove in both circuits but
5.7. SIMULATION 69
after implementation of the basic motor control feedback in Figure 34 the system stabilised. The
circuit shown in Figure 34 is the final stage of the energy recovery system simulation including the
Lynch motors, Maxwell supercapacitor and an approximation of the 4QD motor controller. When
the voltage difference between the previously charged state and cut-off is too low the simulation
will not run. The voltage and current decrease too quickly for the solver to cope. It was found that
a difference of over 4 V was required for the simulation to run. Moreover, where the current had
been high when charging a low voltage supercapacitor, the opposite is true for the discharge. For
simulation purposes the optimum initial voltage previously was found to be 7 V.
Figure 35: The change in voltage of the supercapacitor with respect to the time taken to regenerate
The presence of the earlier resistor has been necessary in order to reduce the current in the
system as attempting alternative motor control encountered simulation errors. However, the voltage
drop across such an element will reduce the power available to the motors. Therefore, the distance
the train will travel under recovered energy should represent a minimum value. It is recommended
that for future simulations an alternative method is found to limit the current of the system using
the motor controller. Figure 36 is the graph displaying the velocity for the locomotive when using
the recovered energy and it also is within competition limits. It is hoped that in actual operation the
train velocity will reach a greater value.
Figure 36: The velocity of the train under energy gained from regenerative braking
The train velocity output data is then integrated and from this method it can be estimated that the
train will travel approximately 3.3 m under only the energy recovered during regenerative braking
and then allowing the locomotive to “free-wheel” (see Figure 37).
70 CHAPTER 5. ENERGY RECOVERY
Figure 34: Circuit diagram detailing the arrangement of components used to simulate the locomo-
tive energy recovery system
5.8. ENERGY RECOVERY SYSTEM TESTING 71
Figure 37: The distance covered by the train using previously recovered energy
The first energy recovery test with the real components was conducted on the bench top between
0 and 5 Volts. The aim of this test was to prove that the components of the circuit were adequate
for use.
To represent the capacitor discharging circuit on the bench top Figure 38 was created. The
components were set up as displayed in Figure 38. Multimeters and clamp meters were used to
measure the voltage across components and the charge and discharge current of the supercapac-
itor.
W
SW80b-5 (Normally Off) Manual Discharge Switch
Energy Dissipation
Resistor 2.2 Ohm
Desk Power Supply V Supercapacitor
Figure 38: Circuit diagram of the energy recovery bench top test
The supercapacitor was charged to 5 V through the soft charge resistor. The charge was held in
the capacitor, checking if the voltage across the supercapacitor did not drop at a rates greater than
1.1 mV per minute. This tested that the supercapacitor could hold charge and the diodes operated
correctly. The energy in the supercapacitor was then dissipated through the 2.2 Ω resistor.
This test proved that all system components were operational and that the system was ready to
test at the full operating voltage of 40 V.
The only circuit modification from the previous test was that the desktop power supply was ex-
changed for the 3 kW DC SMPS. The system is again described by Figure 38.
In addition to the components seen in Figure 38, a thermocouple was used to monitor the resis-
tor temperature, the clamp meter to measure the charge and discharge current, and a multimeter
to monitor the supercapacitor voltage.
The system was set up to charge the supercapacitor through the soft charging resistor. The su-
5.8. ENERGY RECOVERY SYSTEM TESTING 73
percapacitor voltage, charge current and charge resistor temperature were measured. The results
of this test are displayed in Figures 39a and 39b.
(a) Current and voltage plots vs time of the charging (b) Current and voltage plots vs time of the dis-
supercapacitors charging supercapacitor
Figure 39 shows the time scale in which the supercapacitor is charged and discharged. These
results are very similar to the the Maxwell discharge chart profile. The charge time was as pre-
dicted, proving that the current and voltage of the module is predictable.
The temperature of the soft charge resistor (1 Ω) was also monitored during this test. The maxi-
mum temperature of the soft charge resistor was 84 ◦ C. This backs up the predicted calculations in
Section 5.6.4 and gives reason to be aware that the resistors are a hazard when the supercapacitor
is charging.
This was a successful test, the system will be installed on the locomotive without any significant
changes.
6. Brakes
6.1 Introduction DW
Any mechanical system which requires a change in kinetic energy needs both a way to add energy
and a method of absorbing the kinetic energy to slow the system. This energy is typically converted
into either thermal energy through friction, or by using the interaction of a magnetic field.
Modern train brakes are controlled by both the driver and fail-safe control systems within the
train. Following the development of technologies to recapture and store kinetic energy during
braking there will be a decrease in reliance of conventional breaking designs.
The current trend of reduction in locomotive mass has seen examples such as the Intercity
Express Programme weighing between 200 and 500 tonnes [67]. The reduced mass of trains
could revolutionise energy use in the rail industry and reduce wear and tear on braking systems.
However, trains are still required to be slowed and stopped under all conditions, giving rise to the
requirement of a brakes subteam as part of this project.
• All brake systems will be limited to a maximum deceleration rate of 1.3 m/s2 (+-0.15 m/s2 ) on
level track (assuming no trailing load)
• The emergency brake shall provide a deceleration rate of 1.3 m/s2 (+-0.15 m/s2 ) on level track
(assuming no trailing load)
• The emergency brake will operate whenever communication with the remote control unit is
lost, regardless of current locomotive operating state, within 0.5 seconds of disconnection
• The emergency brake will operate in the event that service brakes are released but the pres-
ence of an operator in contact with the remote control unit is not detected for a period longer
than 0.5 seconds
• The remote control unit will incorporate an emergency stop button which shall operate the
emergency brake, regardless of current locomotive operating state, within 0.25 seconds
• An emergency brake button, clearly marked and easily accessible, shall be provided on both
sides of the locomotive which shall operate the emergency brake, regardless of current loco-
motive operating state, within 0.25 seconds
74
6.3. BRAKING FORCE - INITIAL CALCULATIONS 75
• The emergency brake system shall feature a manual brake release to enable movement of
the locomotive by external means when it is safe to do so. Manual reapplication of the brakes
shall also be possible.
• The brake systems on the locomotive shall be so designed that no single fault shall reduce
the available brake effort by more than 50%
• Operation of the emergency brake by any means shall set the traction rate to zero.
Given that the rates of maximum deceleration for both the emergency and service brakes are
equal, it was felt during early stages of the project that a common actuator for the two braking
systems would be preferable in order to simplify the arrangement of the system on the bogie. This
was identified as allowable under the IMechE rules but It was clarified by the competition organisers
that the control systems for the emergency and service brakes should be separate. Consideration
was still given to systems that feature separate actuators in order to retain a variety of suitable
designs. It was also noted that the complexity added to the braking system by the manual release
would have to be fully integrated into the electronics and control systems to fully conform with the
rules.
(a) Pneumatic brakes fitted to a Japanese (b) Eddy current brakes undergoing testing
Bullet Train [72] [73]
Another technology used in modern trains is eddy current braking, which generates a magnetic
field around a permanent magnet on the shaft when the locomotive needs to slow down. The
interaction of the magnet with the field creates eddy currents, impeding the rotation of the axles,
slowing the train [74]. Since the eddy current magnitude is proportional to the rotational velocity of
the axle, brakes of this design cannot be used to bring the train to a complete halt [75], though are
a highly effective method of achieving rapid deceleration [76]. The benefit of eddy current brakes
is the lack of wear on components, although secondary braking systems are still required [77].
Figure 41: The relationship between disc size and force required to overcome its motion
If the disc brake becomes too large, the supplied brake force will not be sufficient. In order
to balance these factors, a brake disc of radius 100 mm was selected. It was considered prefer-
able to minimise the losses due to the mass of the rotating components in order to maximise the
effectiveness of the braking force supplied.
Table 13: Calculations made to assess required force output from any locomotive braking system
The figure of providing 2000 N of stopping force to any brake callipers was therefore taken
forward as a minimum target value when considering the selection of braking systems.
• A vacuum pump: a vacuum pump consists of two one-way valves which surround a small
chamber. Air is forced out of the chamber and the outlet to the air then closed. The inlet is
then opened, causing air to enter the chamber. This process is then repeated many times
until a vacuum is achieved and the process stops.
• Actuators: vacuum brake actuators, like air brake cylinders, have a diaphragm which is forced
to contract upwards when a vacuum is only present on one side of the system. They are
designed such that air can pass through from the upper chamber to the lower chamber,
allowing a vacuum to be created throughout the system before it is purged on the lower side
of the diaphragm. Each cylinder can supply a maximum of 550 N of force at atmospheric
pressure [84] which based on the calculations made in Section 6.6.2 would be more than
sufficient to slow the train if four cylinders were used.
6.6. PROPOSED DESIGNS 79
• Reservoirs: unlike those used in air braking systems, vacuum brake reservoirs do not need
to be protected from depressurisation. Simple, lightweight PVC components exist which can
store the vacuum and limit the effect of leaks on the overall system.
• Valves: a solenoid valve can be used to control the flow of air through a system. When
a current is applied the valve changes position and this in turn can be used to control the
direction of fluid flow through the solenoid. The valve would connect the braking system
to the pump when first starting the braking system, to charge the reservoirs, and when the
operator wishes to release the brakes. In all other circumstances the system will be open to
atmospheric pressure.
• Piping: as with an air braking system, PVC tubing is needed to connect all of the components
together. The volume of the PVC tubing is very small compared to the overall volume of the
components, so the location and length of tubing, within reason, should not have any effect
on the overall performance of the system.
Table 14: Binary weighting matrix showing the relative importance of braking parameters
Based on this assessment, a vacuum brake system was selected to be used within the loco-
motive. Four actuators were to be used, located on the bogies along with the reservoirs and the
pump located inside the frame of the train. Given the compact size of the vacuum pump and the
easy availability of specialist parts specific to miniature railways, design of such a system was
considered relatively simple and sufficiently reliable to be selected.
80 CHAPTER 6. BRAKES
• Cable slackening: if a brake cable or pivoting rod were used to transfer braking force from
the braking cylinder to the actuator, a handle or lever to disable the transfer of force could be
used. The release of the handle would then result in the brakes being reapplied, in a form of
dead man’s handle which prevents train runaway.
• Counter force on callipers: as a second option, an opposing force could be applied to the
callipers, reducing the net force acting on the brake disc as shown in Figure 42a. It would not
be necessary for the brakes to be released completely for movement of the train to occur. A
human male can supply around 130 N of force through their hand when it is in a power grip
position [85]. The use of rods and mechanical advantage would allow one person to provide
sufficient force to counteract the action of the brakes. Alternatively, a ratchet brake similar
to the handbrake in a car was proposed, which could be used to hold the brakes off and the
released, though this was discounted due to the possibility of a runaway train if grip on the
handle was lost.
• Manual operation of the vacuum pump: during ordinary operation, the braking system is
released by creating a vacuum on the lower side of the cylinder by closing off the pipes with a
solenoid valve and the use of an electronic vacuum pump. The same effect could be achieved
by using both a manually closed valve and a hand powered vacuum pump. The premise of
a manual vacuum pump is very similar to that of an air pump for balloons with the valves
reversed. Such systems are used in applications ranging from vacuum sealing in the third
world where electricity is not available to bleeding air from brake fluid in cars.
Due to the design of the bogies, the direct distance between the brake cylinder and the actuator
was limited to around 20 mm. Due to the constraints of positioning and sizing, it was decided that
cable slackening brake cables or brake rods was not an ideal system to be used in the locomotive.
6.7. INTEGRATION WITH OTHER SYSTEMS 81
It was also decided that loading both the callipers and the cylinder with an additional counterforce
could be damaging to the diaphragm within the brake cylinder. Therefore, the manual vacuum
pump was selected as the best method for applying vacuum brakes.
Figure 43: The solenoid valve used to control the braking system
The selection of the brake solenoid is discussed in more detail in Section 7.4.3. Although it was
primarily selected for its ability to handle vacuum pressures, the volumetric capacity and its size
compatibility with the vacuum brake components. The overall simplicity of the system means that
the emergency systems needed to apply the brakes in event of a detected fault or the press of the
emergency stop button can simply cut power to the valve in order to guarantee maximum braking
force.
The system was also fitted with pressure sensors in order to monitor the performance of the
braking system. These will allow the operator to guarantee without visual inspection that the brakes
are functioning as expected and that leaks have not occurred that will impair the operation of the
system.
The flow of air around the system is also displayed in Figure 44. Arrowheads indicate the
direction of air flow during normal operation, and dotted lines indicate areas which are controlled
when the manual release is being used. Finally, green lines indicate areas that are permanent
vacuums during normal use, and yellow lines vacuums that are temporary depending on whether
the brakes are applied or not.
Originally, it had been anticipated that either a cable or direct rod connection could have been used
to connect the vacuum braking system to the calliper. However, the actuator travels in a purely
horizontal movement and the brake callipers move in an arc pattern, meaning that a rod design
was discounted. A system was devised, shown in Figure 45, connecting the two components but
also allowing a small amount of rotation in the plate connection between two bolts. The plates are
both made of 4 mm thick steel. The system is held in place by a number of hex bolts.
In order to confirm that the design was suitable for use within the locomotive, it underwent static
6.8. FINAL DESIGN 83
stress testing using an FEA package in SolidWorks. The braking system was modelled by applying
a pull force to the cylinder brake rod and fixing the bolt that attaches to the callipers. The results
are shown in Figure 45 and confirm that the method of interaction is suitable. The factor of safety
at its lowest point is 1.5, but the majority of the system has a factor of safety of at least 5. Given
that the system will be under less stress in service, due to the movement of the brake calliper lever,
this factor of safety is allowable.
6.8.3 Callipers
The brake callipers, seen in Figure 47a were sourced from an American company, “Air heart brake”,
selected for their size and ability to handle the 550 N of force supplied by each actuator. The cal-
lipers were designed using imperial measurements, meaning that in order to allow proper inter-
action between the actuator and the callipers the holes on the brake lever were drilled out to M8
holes.
(a) A side view of the Airheart MB1 (b) The completed brake disc attached to
Brake Caliper the wheelset
Figure 47: Two components forming the connection between the braking system and the track
The table shows that the projected rate of deceleration is just over 1.22 m/s2 , which is within the
required tolerances set by the IMechE. In order to reduce the overall reliance on maximum capacity
operation of the vacuum pump while still maintaining the required performance, investigation into
enhancing the coefficient of friction of the brake pads was set as a pre-competition task.
voirs and emergency release valves as a return to atmospheric pressure renders the brakes non-
functional. The manufacturer of the vacuum reservoirs recommend sealing of the components with
epoxy glue but in addition tape was placed around the edges of the cylinder to protect the seal from
damage and for aesthetic reasons.
6.11 Testing DW
In order to confirm that the brakes were assembled correctly and performed as expected, a number
of tests were carried out before the braking system was fitted into the locomotive. Initially, these
concerned only the function of the vacuum braking system rather than callipers and brake disc due
to delays in the arrival of the brake callipers. Full systems testing including the brake callipers is
discussed in Section 13.
This set of instructions was followed multiple times throughout the testing period in order to
confirm that the system functioned as expected. The system was confirmed to work through mea-
surement of the distance between the brake rod tip and the brake casing. The rod can be seen
moving to its release position as the handle is squeezed and returning to a fully applied position
when the valve was reopened. The total travel of the rod was 20 mm.
(a) Leak testing and detection with one (b) Operational testing with three brake
brake cylinder cylinders
7.2 Microprocessor QA
7.2.1 Introduction
In the 1940s, Integrated Circuit (IC) technology was born due to the desire to produce improved
military technology during the Second World War. ICs offered the advantage of harnessing hun-
dreds of components in a very small area in addition to the ability to increase systems complexity
and integration. With the invention of Metal Oxide Semiconductors (MOS) in 1971, IC design then
evolved into microprocessors allowing greater performance and reliability. With the rise of Mecha-
tronics as a result of these advancements, many applications nowadays rely on the integration
of electronic technology and mechanical systems using microprocessors as tools to control and
process information and make decisions [86].
A microprocessor is essentially a small computer featuring three main components:
Microprocessors are powerful tools that are able to control various processes such as industrial
87
88 CHAPTER 7. ELECTRONICS AND CONTROL
automation, electric current, temperature and engine performance. Nowadays, the application of
microprocessors in railway systems allows the control of various systems in the locomotives as well
as monitor on-board equipment. Modern railway trains such as the Japanese Shinkasen, British
High Speed Trains (HST) or the French TGV (Train à Grande Vitesse) utilise microprocessors to
control safe and reliable traction systems that enable the trains to go beyond speeds of 200 mph
[87].
One of the main requirements of the IMechE Railway Challenge is to be able to control all
functions on the locomotive using a radio-based or a cable-based transmission. This meant that
full electronic control of all functions and systems must be used as part of the locomotive design.
Due to the number of functions required to be controlled, a microprocessor was required.
The first parameter to consider in choosing a microprocessor is its input and output capabilities.
This includes the type of IOs the microcontroller can cope with such as analogue and digital signals
and the number of corresponding pins available. The choice is therefore dependant on the functions
required from the application. At an early stage of the design phase and with rough knowledge of
the functions required, a microcontroller with a large and broad range of IOs was required.
Cost
The cost of microprocessors, much like computers, varies depending on the features they offer.
Some microprocessors can be as cheap as £25 but are very slow and can only process digital
signals whilst others can cost hundreds of pounds but can perform the tasks a normal personal
computer would.
Speed
Speed is an important technical parameter to consider. Although most microprocessors can meet
most applications’ speed requirements, the speed can reduce as large data volumes and infor-
mation are fed into it. In a lot of applications, this is avoided by using multiple microprocessors
performing different tasks at the same time.
As mentioned earlier, the rules of the competition require that all functions in the locomotive are
operated remotely via a radio-based or a cable-based transmission. To allow remote use but not
risk reliability, a microprocessor with both types of transmission was desired.
Software
The software tools used to program a microprocessor is a major consideration in choosing one.
The programming languages vary between microprocessors and the programming experience of
the group was an important consideration. It was essential to select a microprocessor that has an
easily programmable software that allows flexibility and makes the process of debugging easy and
quick.
7.2. MICROPROCESSOR 89
Memory
Arduino boards have become the largest and fastest growing hardware and software embedded
platforms and has become extremely popular in the robotics hobbyists market. Arduino boards
are good solutions to perform simple tasks from flashing LEDs to moving a small aircraft wing.
They are based on C++ programming experience although can be implemented using LabVIEW.
Arduino boards vary in price but are considered to be relatively cheap with prices ranging from
£30 to £180. They have a large online platform with hundreds of sample codes available on online
forums and communities. With 14 digital IO pins and 6 analogue input pins, Arduino boards can
be limited to prototyping but cannot operate an entire full-scale system. They also have a relatively
slow processor with a maximum speed of 20 MHz and a small memory that can only cope with very
few megabytes of data. Furthermore, Arduinos come with an Ethernet cable to communicate with
their software tools. An add-on shield is required to be able to use wireless communication such
as Wi-Fi or Bluetooth and these can significantly reduce the speed of the board when used [88].
Raspberry Pi
Raspberry Pi single-board computers were introduced into the market in 2012 targeting young gen-
erations and encouraging them to enter computer science disciplines. A complete functional mini-
computer, the Raspberry Pi can be programmed using a wide range of programming languages
such as Python, C/C++ and Bash which gives the opportunity for a broad range of programming
experiences. Due to its Linux operating system and the need to install code libraries, Raspberry
Pi coding is not as flexible as other microprocessors. On the other hand, it is 40 times faster than
Arduino boards and has 128,000 times more memory. Raspberry Pi can multitask processes mak-
ing it an excellent tool for projects requiring more interactivity and processing power. At a relatively
cheap price of about £30, Raspberry Pi relies on an Ethernet cable to communicate with the soft-
ware. With only 8 IO pins, the mini computer is limited when it comes to reading analogue signals
as it does not have an onboard analogue-to-digital converter (ADC) [89].
Similar to the Raspberry Pi BeagleBone borads are mini computers with analogue inputs. They are
generally more expensive than other microprocessors at a price range of £85-£110. They contain
a 1 MHz processor and 69 IO pins offering the fastest and most comprehensive IO capability.
BeagleBone boards are programmed using the same Linux operating system that Rasberry Pi
uses making the software experience difficult [90].
The myRIO microprocessor is the most recent addition to NI’s Reconfigurable IO (RIO) architecture
line of controllers. It is widely considered the most student-friendly microprocessor on the market.
With LabVIEW as the main programming language, the graphical programming language offers an
easy and an interactive programming environment for designing and integrating complex systems.
The myRIO includes ten analogue inputs, six analogue outputs, forty lines of digital IOs in addition
90 CHAPTER 7. ELECTRONICS AND CONTROL
to two audio channels. It also has an onboard Wi-Fi hotspot enabling easy communication with
laptops or tablet devices without the need to use Ethernet cables. The Field Programmable Gate
Array (FPGA) in the internal architecture gives the microprocessor high core speed as well as the
flexibility to prototype and iterate designs quickly. Although expensive, the comprehensive tools
that the myRIO provides give a reliable platform to operate and integrate systems in an easy and
quick manner [91].
Microprocessor GPIO pins Hardware and Software Communication Cost Core Speed Software Memory
Arduino Boards 14 Ethernet Cable £30 16 MHz Ardunio IDE (C/C++) 64 MB
Rasberry Pi 8 Ethernet Cable, Wi-Fi, Bluetooth £30 700 MHz Python, C/C++, Bash 512 MB
BeagleBone Black 69 Ethernet Cable, Wi-Fi, Bluetooth £100 1 GHz Python, C/C++, Bash 512 MB
NI myRIO 40 Ethernet Cable, Wi-Fi, Bluetooth £500 667 MHz LabVIEW, C/C++ 512 MB
Figure 51: Binary weighting matrix showing the relative importance of microprocessor parameters
Having done that, a score on a scale of 1-6 was given to each of the microprocessors based on
the parameters presented in Table 16. As can be seen in Figure 51, the NI myRIO had the highest
score and was therefore chosen as the microprocessor to be used in the project.
Table 17: Subsystem functions and the corresponding input and output requirements from the
myRIO microprocessor
For each of the functions, LabVIEW Virtual Instruments (VIs) were created and tested with the
92 CHAPTER 7. ELECTRONICS AND CONTROL
required hardware as will be shown in the following sections. Within each VI, global variables were
created and added to a tablet variables library. Using the Wi-Fi hotspot feature on the myRIO
microprocessor and the NI Data Dashboard application, the global variables can be used to create
a user-friendly interface to control the functions remotely from a tablet as shown in Figure 53.
Measurement Methods
• Optical Sensors
Optical-based methods of speed measurement rely on the principle of photodetection whereby
an LED is aligned on one side of a perforated opaque disc as shown in Figure 54. On the
other side of the disc, a photodetector is fixed opposite to the LED light. When the disc
rotates, the LED light alternately passes through the transparent parts of the disc. The varia-
tion of the light output produces a square wave signal of which the frequency can be used to
calculate the speed of the wheel or the vehicle based on the number of holes on the disc.
The main advantage of using optical sensors is that they provide a contactless measurement
method. They are known for being accurate and sensitive. Although they are used in a wide
variety of industrial applications, optical speed sensors have a number of weaknesses. The
use of optical components means that sensors are always susceptible to dirt and interfer-
ences from other sources. Traces of dirt or fog can significantly reduce the amount of light
passing to the detector causing unreliable readings and measurements. Another problem
with these type of sensors is the ageing of the LEDs. With their short lifetime, LED light
signals can decrease or deteriorate quickly causing another reliability problem. With unpre-
dictable weather conditions on the track, the lack of reliability as well as the cost implications
of making a perforated disc, optical speed sensors were discounted.
• Global Positioning System (GPS)
Speed measurement using GPS are based on a series of trackpoints that record the posi-
tion estimates (latitude and longitude) of an object that are taken at regular time intervals.
The distances between the consecutive positions and time values are then used to calcu-
late the speed [93]. Although seemingly a simple method, GPS speed measurement was not
considered further because it can be unreliable and inaccurate. Each trackpoint is usually as-
sociated with an error due to satellite signals delays. Therefore, the accumulated measured
distances and hence the calculated various speed can be misleading.
• Accelerometers
Accelerometers are very powerful rugged devices used in a wide range of applications to
measure acceleration, vibration, position and orientation. The fundamental mathematical
principle behind using accelerometers to measure speed is to integrate the acceleration sig-
nal to obtain speed. However, the signal processing of accelerometers is not as simple as
that. As an object moves and varies its speed, the signal level varies from its first initial value.
This introduces an error into the velocity measurement known as drift which is equal to the
amount of variation. Integration as an addition process will keep adding the errors leading
to inaccurate speed values. This can be avoided by installing more than one accelerometer
but the signal processing remains difficult and therefore accelerometers were not considered
further for speed measurement.
• Hall Effect Sensors
Much like optical methods, Hall effect speed measurement is a contactless method. Based
on the principle that a sensor varies its output voltage due to a magnetic field, Hall effect
sensors are very popular as wheel speed sensors due to their accuracy and reliability. Usually
Hall effect devices are configured as on-off switches giving a high state when the sensor
passes by a magnetic field and a low state when the sensor is not aligned with the magnet.
Similar to optical methods, Hall effect sensors generate a square wave which can be used
to calculate the speed using the time intervals between the high states [94]. This method
offers flexibility as to where the source of the magnetic field is. The magnet can be inside the
sensor and a high state is induced when a ferromagnetic material aligns with the sensor. On
the other hand, the source of the magnetic field can be an external permanent magnet with
the sensor producing a high state every time it aligns with the magnet. With this amount of
flexibility, accuracy, and reliability, Hall effect speed measurement was considered to be the
most effective method for calculating the speed of the locomotive and hence was chosen to
be implemented.
94 CHAPTER 7. ELECTRONICS AND CONTROL
To allow for design adaptability, the NPN sensor shown in Figure 55a was chosen. The bolt-like
design gives the flexibility to freely vary the distance between the magnet and the sensor. As
seen in Figure 55b, the sensor has a built-in magnet which means that an output voltage can be
produced by a ferromagnetic material or an external permanent magnet. The sensor is powered
using the 5 V DC supply pin from the myRIO microprocessor with a 1 kΩ pull-up resistor between
the output and the supplied voltage to ensure a positive output (high state) is produced from the
sensor when it is aligned with a magnet.
To test the sensor, the LabVIEW VI shown in Figure 56 was created. A timer starts when
a high state is generated due to the alignment of a magnetic field with the sensor. The time
difference between two consecutive high states is recorded and is inverted to obtain a frequency.
The frequency value can be used to calculate the speed of a rotating part by using:
C
Speed (m/s) = RPM (7.3.1)
60
where C is the circumference of the rotating part.
The sensor was tested with the Lynch traction motors with the set-up shown in Figure 57a.
A magnet was attached to the motor shaft with a 1.5 cm distance from the sensor; sufficient to
7.3. SENSOR SYSTEMS 95
produce an output voltage. Using the LabVIEW VI developed to control the motor as shown in
Figure 57a, the voltage across the motor terminals was varied, rotating the shaft and producing
both an RPM and a speed indicator using the front panel shown in Figure 57b.
Whilst testing, one problem that was noticed with this methodology is that the timer is contin-
uously counting after a high state. This means that when the rotating part stops, the speed value
does not go to zero as the last time value is still recorded. To overcome this problem, the MATLAB
code shown in Figure 58 was written. The code defines a threshold time value and controls whether
the timer continues or stops. If the timer value is lower than the set threshold, the counter should
continue and a speed value is calculated. If the threshold value is exceeded, the timer should stop
and a zero speed value is produced. The threshold time value can be determined by the slowest
expected speed of the rotating part.
After evaluating space in the bogie assembly, it was decided to use the axle as the rotating part
for measuring the locomotive speed. A magnet is put in the bracket shown in Figure 59a which
is fitted to the axle. The sensor is mounted on the L-shaped bracket as seen in Figure 59c. Both
96 CHAPTER 7. ELECTRONICS AND CONTROL
brackets were manufactured using 3D printing techniques using PLA plastic as the material as
seen in Figures 59b and 59d.
(c) Sensor bracket modelled in CAD (d) A prototype of the sensor bracket
To achieve these requirements, research was conducted on the different types of heat detection
systems to determine their suitability and application. In general, there are two methods for detect-
ing fire from the presence of heat:
7.3. SENSOR SYSTEMS 97
• Fixed Temperature Heat Detectors which operate when the measured temperature exceeds
a predetermined temperature.
• Rate-of-Rise (ROR) Heat Detectors which operate when the measured temperature in-
crease over time is equal or greater than the rate of change the detector was manufactured
to operate at.
Measurement Methods
• Electromehcanical
Electromechanical heat detectors operate due to a mechanical movement that sends an elec-
trical signal. The most common type is a bi-metallic strip which is has one end fixed whilst
the other is free to move depending on its temperature. When the temperature is increased,
the bi-metallic strip expands and completes an electrical circuit that actuates an alarm.
• Optomechanical
Optomechanical detectors contain one or more fibre optic cables separated by a heat sensi-
tive insulator and protected by an outer case. A light signal is passed through the fibre optic
cable so that when the detector is exposed to heat, the insulator melts. This implies that the
light signal will be discontinued while activating the fire alarm.
• Electronic (Thermistors)
The most common type of heat detectors are those that use thermistors as their heat sensitive
element. Thermistors are resistors whose resistance changes according to temperature.
The resistance-temperature relationship is non-linear and therefore thermistors can be very
sensitive to temperature changes making them a good candidate for fire detection.
By monitoring the voltage across the thermistor, the resistance can be calculated using the po-
tential divider equation. The value of the resistance can then be used to calculate the temperature
98 CHAPTER 7. ELECTRONICS AND CONTROL
RT
Vout = Vin (7.3.2)
RT + R1
R1 Vout
RT = (7.3.3)
Vin − Vout
β
T = (7.3.4)
ln( rR∞T )
−β
where r∞ = Ro e To , Ro is the resistance measured at room temperature, β is the thermistor coeffi-
cient and To is room temperature (i.e 25o C).
To test the system, the circuit was built on a breadboard as shown in Figure 61 with a heat gun
as the main source of heat for testing.
The VI shown in Figure 62 was created to achieve the three main system requirements. A
threshold value can be set depending on the temperature conditions inside the shell of the locomo-
tive. When that value is exceeded, a digital signal is sent to the brakes channel to activate them.
Simultaneously, the digital signal activates a pop-up window to inform the operator of the fire. With
regard to the sound alarm, this can be achieved by using an analogue output on the myRIO with a
sine waveform generator.
disc allowing the locomotive to move and when the vacuum is released the brakes are automatically
applied. The control of the braking system involves manipulation of the pressure inside the braking
system.
• Maximum deceleration of 1.3 m/s2 : The pressure must be manipulated inside the system
such that the force applied to the calliper does not cause a braking rate above specification
deceleration.
• Automatic application of the brakes in the event of a power failure and in case remote control
is lost: the pressure manipulation device has to be fail safe, such that when power is lost the
system is re-pressurised to atmospheric pressure and the brakes are applied. The myRIO will
constantly monitor the status of the connection with the control device (tablet). If connection
is lost the brakes must be applied.
• Emergency stop button (apply brakes within 0.25 seconds): an emergency stop button will
activate the brakes by re-pressurising the system
• Indications of braking fault: pressure sensors are required to monitor the status of the vac-
uum.
Taking these rules into consideration greatly influenced the system design, giving clear deliver-
ables to work towards.
• Automatic internal pressure triggered switch for shut off when the system is evacuated, re-
ducing energy consumption
• Adjustable pressure switch for manual adjustment in baseline system pressure, setting the
maximum braking force
• Internal fusing and circuitry protection.
To vary the braking force applied to the brake callipers, a method of electronically manipulating
the pressure inside the braking system was necessary. Vacuum brakes traditionally have been
used with manual progressive valves to vary the pressure inside the system. To enable the remote
control of this system, electronic variants of these valves were required. Progressive electronic
valves are not readily available with the majority of electronically controlled flow devices being
on/off solenoid valves. The selection of this solenoid valve depended on a number of criteria:
• Integration with the components already selected. All connections have to fit with the rest of
the system.
• The valve is able to operate with the working fluid as air
100 CHAPTER 7. ELECTRONICS AND CONTROL
• A suitable operating voltage and current for integration with rest of the control system
• The flow rate through the valve is enough to satisfy the safety cut-out response time.
The first point was easily achievable by specifying that the solenoid (through connection adapters)
could mate with the PVC hose supplied by PNP railways. The majority of valves on the market are
capable of working with a number of working fluids, air being the most common. A useful solenoid
coil operating voltage would be 12 V, as the majority of subsystems have components operate at
this voltage. This is not a common voltage for the size of the valve required but a simple DC-DC
converter or battery packs in series could be used. A market search of solenoid valves yielded
a suitable manufacturer with a wide product range, SMC solenoid valves. A threaded brass body
allows for barb fittings to be attached, simplifying the pipe-valve interface. To narrow down the SMC
product range, it was useful to review the required response time. This is the most important factor
as it will give an indication of the latency between a control input and a mechanical response. A
simple volumetric estimate of the braking system, coupled with the necessary response time from
the rules gave a required flow rate:
Using the SMC datasheet [96] the volumetric flow rate can be approximated as a function of the
upstream and downstream pressures (P1 = 0.1 MPa and P2 =0.02 − 0.03 MPa. This is illustrated by
the red line in Figure 63, showing a flow rate of around 30 L/min. By comparing the required volu-
metric flow rate with the value achievable with each series of SMC solenoid valves, an appropriate
valve was be selected. The VX311 series provides more than adequate flow rate for the required
response time. Some assumptions were made in performing this calculation:
• Flow rate will not be constant throughout the re-pressurisation of the braking system, as P1
and P2 change over time
• The brakes may be ‘applied’ despite the system not being fully evacuated, full application
occurs when the system is at atmospheric pressure
• The calculation does not take into account the pressure differential across the membrane
inside the actuator. This will increase the flow rate through the solenoid valve by changing
the controlled volume, maintaining the upstream pressure for a prolonged period of time.
Despite these assumptions the analysis gives a positive indication that the braking system will
perform to the required specification.
7.4. BRAKE SYSTEM 101
Figure 63: Upstream and downstream pressure against volumetric flow rate [96]
Once this solenoid had been selected, the next step was to integrate with the myRIO microcon-
troller.
DC
5V Digital Input
Figure 64: Circuit diagram for the standard transistor switching circuit
102 CHAPTER 7. ELECTRONICS AND CONTROL
The simple circuit seen in Figure 64 can be used for switching any high powered devices and
will also be used in subsequent sections for the motor controller and energy recovery circuitry.
A digital output from the myRIO can be connected to the base pin of the transistor to switch the
solenoid valve on and off, with a simple VI to operate it as seen in Figure 65. To enable the tablet
to operate the solenoid valve, a network shared variable called ‘On Off Switch’ was created in a
variable library within the myRIO. These variables can be called into multiple VIs at the same time
as they are global variables.
• Leaks - The constant escape of air from the various joints and connections made the pump
continuously switch on and off. This was reduced by using PTFE on threaded joints and
taping all barb-pipe junctions.
• Pump hunting - The pump continuously switched on and off for very small changes in system
pressure. This could cause excessive wear on the pump so alteration of the internal pressure
switch was conducted to reduce this.
These promising results and control system improvements from a control aspect gave confidence
for the future full system integration.
impact the rules have on the design requirements, component selection, and construct the required
circuitry around the components.
• The locomotive must be able to be controlled via radio or cable based transmission: Motor
controller must be able to be operated remotely
• Maximum speed is 15 km/h: Motor control system must incorporate speed feedback with
appropriate algorithms in place to limit speed
• Maximum deceleration rate using energy recovery is 1.3 m/s2 : Motor controller deceleration
rate must be able to be modified. Energy recovery must be possible with motor controller.
These rules gave a set of features and design parameters to begin the component selection. Fur-
ther to these rules, calculations in the Section 4 provided additional parameters to aid the com-
ponent search. A maximum current of 200 A is required at a maximum voltage of 48 V DC. This
reduced the number of controllers greatly as they are not as readily available as 12 or 24 V con-
trollers. There are additional features which were identified early on as being desirable for refining
the locomotive design. One example is acceleration/deceleration ramp control which dictates how
quickly the locomotive accelerates to a desired input speed. This is useful for reducing the chance
of wheel slip under extreme track conditions or trailing load. An option for freewheeling (disableing
regenerative braking) is a desirable feature. This would be used in the energy recovery challenge to
maximise the distance travelled from the amount of energy recovered by freewheeling the motors,
coasting the locomotive when the recovered energy has been consumed.
Percentage Weighting
Offset Product Score
Acceleration Ramp
Customer Support
Controller Display
Forward/reverse
Energy Recovery
Fault detection
Free Wheeling
Product Score
Cost
Energy Recovery X 0 1 0 0 0 0 0 0 1 2 9.09%
Free Wheeling X 1 0 0 1 0 1 0 3 4 18.18%
Forward/reverse X 0 0 0 0 1 0 1 2 9.09%
Cost X 1 1 0 1 0 3 4 18.18%
Customer Support X 1 0 1 0 2 3 13.64%
Acceleration Ramp X 0 1 0 1 2 9.09%
Fault detection X 1 1 2 3 13.64%
Remote Control Support X 0 0 1 4.55%
Controller Display X 0 1 4.55%
Total 22
Binary Input 1 0
Blank Cells X
As can be seen in Figure 66 there are some clear shortfalls in certain controllers. The Alltrax
controller does not have internal reversing functions, it requires external switching circuity hence
it scored zero. The 4QD does not provide a controller display, but as this is classified as a low
percentage importance when compared to other design requirements, it does not impact too much
on the product design requirements satisfaction percentage. The selected motor controller from
the binary weighting matrix is the Sigmadrive PMT465L, where the main positive points are its ad-
justability through a hand held programmer, the fault detection capability and the display. During
the testing phase, the selected motor controller was successfully spinning the motors by using a
0-5 V output from the myRIO until a fault developed limiting the operation of the motor controller.
Due to the point in the project this failure occurred, sourcing another Sigmadrive unit was challeng-
ing due to the remaining time available and the long lead times. The only option was to source
another controller, so analysing the binary weighted matrix again, the 4QD was selected as the
final controller.
Battery Terminals
B-‐ B+
1N4148
myRIO 100<ё
Contactor SW-‐280
A B C D
LED
4 Way Connector
1<ё
B-‐ B+
10<ё
Controller Terminals
A B C D E F
6 Way Connector
The controller circuit boards contain Insulation Displacement Connectors (IDCs) seen as the 6
way and 4 way connector in Figure 67. These allow control signals, external controller indications
and safety features to be connected. Starting with the 4 way connector, pins A and B operate the
pre-charge resistor contactor bypass. The 470 Ω resistor pre-charges the internal capacitors of
the motor controller before a direct connection to the supercapacitor terminals is created. The key
switch between C and D enables the controller to operate and the contactor to close. Moving to
the 6 way connector, the LED between pins B and F shows the controller is operational and only
displays when the key switch is closed. The transistor state between pins C and F decides whether
the motors spin in forwards or reverse, and pin C is pulled up to battery positive. This transistor
is operated by a digital output from the myRIO, operated in a similar way to the VI illustrated in
Figure 65. The 10 kOmega resistor between pins D and F is a pot defeat resistor, mimicking the
variable resistor in a manual throttle that would be connected to pins D, E and F. This enables pin
E to connected to a 0-5 V output from the myRIO, which acts as the throttle.
As discussed previously, to maximise the distance travelled during the energy recovery chal-
lenge, the regenerative braking feature would be useful to disable. This mode of operation is not
one that is usually used by electric vehicles as it disables a system designed to increase efficiency
by regenerating energy during the braking phase. The motor controllers investigated previously
are all four quadrant controllers, having four modes of operation – drive and regeneration in both
forwards and reverse. Figure 68 shows how quadrant one (Q1) and Q3 accelerate the motor in
forward and reverse and Q2/Q4 decelerate the motor in forward and reverse.
106 CHAPTER 7. ELECTRONICS AND CONTROL
With the 4QD, this four quadrant operation can be defeated by completing a simple board mod-
ification. Two contact patches on the top board of the controller were connected to a switch which
disabled or enabled the regenerative braking. This has the impact of disabling quadrants two and
four, creating a two quadrant controller. The torque applied (during regenerative braking) to the ro-
tor in the opposite direction of rotation now is not present. This will enable the locomotive to travel
as far as possible using the energy recovered during the braking phase of the energy recovery
challenge. During normal operation the controller will operate in four quadrant mode as designed.
Reflecting the more complex circuitry in Figure 67, the LabVIEW application required to operate
the circuit grew in complexity as well. To operate the motor controller a 0-5 V analogue output and
a digital output is required. To manipulate the global variables throttle and reverse in Figure 69b the
National Instruments Data Dashboard application was used on a tablet. A simple graphical user
interface (GUI) was generated to manipulate the 0-5 V output seen in Figure 69a. To safely operate
the train, the VI was designed such that the motors and brakes cannot be applied simultaneously.
The throttle input on the tablet ranges from -5 to 5. Forwards operation is from 0 to 5, a dead-band
from 0 to -1 is where the brakes are applied and -1 to -5 is where the locomotive is in reverse.
7.5. MOTOR CONTROL SYSTEM 107
During the testing of the motor control system, several points were raised regarding the perfor-
mance of the whole system:
• Contactor not closing in horizontal orientation – The contactor has heavy, large contacting
plates requiring relatively large force to move them. In the vertical position, closing upwards,
the contactor would not close. Rotating the contactor such that it closed towards the ground
reduced the force required to move the plates, allowing the contactor to close.
• Motor speed did not correspond to the 0-5 V input - During testing it was noted that the motors
did not start to spin until the voltage output was around 1.8 V. Modification to the gain setting
on the motor controller overcame this, with linear response in motor speed to input voltage,
cutting in at about 0.48 V.
• Motor kick during initial acceleration – When rotating the motors from a standstill a significant
kick was witnessed before smooth acceleration. This is undesirable as a jerking motion from
the motors could cause excessive chain wear/stretch as well as noise/vibration problems.
Conducting the gain adjustment as before minimised this.
• Imbalanced motor speed – When using the motors in series it was noticed that the rotation
speed did not match up. This is due to the minute differences in the loading applied to the
motor through internal resistance and stiction of the bearings. The difference in the load
leads to an imbalance in the voltage across the motors, causing a speed difference. This is
demonstrated in Figure 70 by the different values displayed on the multimeters on the far right
which are connected to each motor.
• Regenerative brake defeat switch – A noticeable increase in spool down time was witnessed
with the regenerative braking defeat switch applied. At no load, the time from an applied 10 V
to 0 V across the motors increased from two to four seconds.
Moving forward from these successful bench tests of the motor control, the next stage was
to integrate the energy recovery system with the supercapacitor. This was completed by using
the switch mode power supply plugged into a wall outlet to charge the supercapacitor, which then
replaced the lead acid batteries previously used for bench testing. A more detailed analysis of this
is found in Section 7.8.
7.6. ANCILLARY BATTERY SYSTEM 109
• To ensure full functionality of the microprocessor and control system, and maintain control of
the braking system in the event of a generator or energy recovery failure
• To reduce the load on the generator during normal operation.
It became clear during the design stage that this was in fact a complex and vital subsystem
in the normal operation of the locomotive. The main components to be powered by the ancillary
system and their voltage and maximum current ratings can be found in Table 18, note that the
current draw during normal operation is expected to be less than the stated values.
of time. Since both a 12 V and 24 V supply were required; the initial design consisted of using
two 12 V batteries connected in series, where components that required 24 V could be powered by
both batteries and then the other components could be split between the two individual batteries in
an attempt to try and keep the current drain on both batteries equal.
There has been a large amount of research into the use of batteries as power sources for electric
vehicles. There are multiple battery types that are suitable for vehicle power sources: lead-acid;
nickel-metal hydride (Ni-MH); lithium-ion (Li-ion); metal-air and sodium-nickel chloride [98]. In order
to decide upon the most suitable chemistry type, it was important to first decide upon the criteria
that the batteries must meet.
The first major point to note is that since a higher locomotive mass corresponds to increased
traction, the energy density of the battery was not of concern. In contrast, safe and reliable oper-
ation were paramount. Since the batteries were also not going to be the main source of power on
the locomotive, it was agreed to try and minimise the system cost and circuit design complexity. Ini-
tial research showed that the three most commonly available battery types for consumer purchase
suitable for use on the locomotive were lead-acid; nickel-metal hydride and lithium-ion. The main
performance characteristics are briefly summarised in Table 19.
The chosen battery technology for primary power in the ancillary system was lead-acid. The
lead-acid battery has been the most popular battery technology for over 100 years [99]. They
have also been used for multiple different applications such as electric and hybrid vehicles; energy
storage; emergency power and as a power source for mining and material handling equipment
among others. Due to the high demand and various usage, they are associated with a low price
and ease of manufacture [99]. However, they do have some limitations: they have large space
requirements and have a high mass due to the use of lead [100]. Their performance also greatly
reduces if they are discharged below 30%. In order to avoid this, a simple battery monitoring system
was implemented where the NI myRIO microprocessor was used to monitor battery voltage and
SOC.
The main reasons for not using the higher performance Ni-MH or Li-ion batteries are due to
the more complicated battery management systems that would be required; both of these types of
7.6. ANCILLARY BATTERY SYSTEM 111
batteries would also need temperature management as well as voltage and charge monitoring. For
example, Li-ion batteries suffer from a degradation of performance when discharged below 2 V [99]
and a permanent loss in capacity at temperatures above 65°C. Not to mention the most obvious
disadvantage, their high cost ($600/kWh) [100].
Table 18 highlights the power requirements of the ancillary circuit; there is an estimated max-
imum current draw of 14.8 A split between two batteries. A desired running time of one day (7-
8 hours) requires a battery capacity between 50 and 60 Ah using using Peukert’s Law, Equa-
tion 7.6.1 [101]:
Cp = I k t (7.6.1)
Battery Charger
As can be seen in Appendix E, there are recommended charging characteristics in order to max-
imise and maintain the performance of the batteries. Two battery chargers that perform a pre-set
charging algorithm were bought at the same time as the batteries. The reason for choosing to pur-
chase battery chargers rather than design and build ones specifically for the batteries was simply
to ensure that the batteries charge safely. There is scope to improve this feature of the train, as will
be discussed in Section 14.
As previously mentioned, the NI myRIO will be used to monitor the voltage and SOC of the batteries
during operation. The myRIO itself does not have the capability to carry out such procedures.
Therefore, two additional pieces of equipment are required, these are:
• Low voltage disconnect - breaks the external circuit when it senses that the battery voltage is
too low
• Current Sensor - By measuring the current, and integrating this value with respect to time
using the myRIO, the SOC of the battery can be estimated.
The integration between the external components and the myRIO will be discussed in Sec-
tion 7.6.5. The low voltage disconnect chosen has an input voltage range between 9 and 32 Vdc
at a continuous maximum current of 20 A. An open loop current sensor was used which consists
of a Hall effect sensor mounted in the air gap of a magnetic core. The current carrying wire that
is surrounded by the sensor produces a magnetic field that is proportional to the current; this field
is measured by the sensor and a corresponding voltage output is supplied [102]. The FW Bell
NA-25 has a rated input current of 7 mA and a sensitivity of 0.9 to 2.5 mV/A, with a linear perfor-
mance relationship up to 100 A. It is expected that the sensor itself will be powered by the myRIO.
This allows the capability to be able to control when the current is being sensed. The low voltage
disconnect and a diagram of the current sensor can be seen in Figures 71a and 71b respectively.
112 CHAPTER 7. ELECTRONICS AND CONTROL
(a) PowerTector PT20 low voltage discon- (b) Diagram of operation of the Hall effect
nect [103] current sensor [102]
This circuit will be implemented using two fused distribution boards (one connected to each
battery) so as to simplify the wiring process of final implementation. The most complex part of the
ancillary system will be including the battery monitoring (detailed circuits can be found in Figures 73
and 77) due to the integration of the circuit components with the myRIO. In order to ensure effective
and accurate operation of the battery monitoring system a series of bench top tests were carried
out, detailed in Section 7.6.4.
7.6.4 Testing
Low Voltage Disconnect
Two low voltage disconnect components were used, one in series with each of the batteries. They
are solid state devices that monitor the source voltage and break the circuit when the voltage falls
7.6. ANCILLARY BATTERY SYSTEM 113
below a pre-determined level. The PowerTector low voltage disconnect units used have a range of
pre-programmed voltage cut-off settings that allow for use in a variety of scenarios. The 20 A rated
unit used, as seen in Figure 71a, can be connected inline and does not require chassis mounting.
The circuitry that will be used can be seen in Figure 73.
Figure 73: Low voltage disconnect circuit, the unit is the component marked Z. Note the use of a
voltage divider circuit to allow for myRIO integration.
Before testing the integrated circuit, first the simple voltage divider circuit was used to help
configure the myRIO. Using a breadboard, the circuit was first tested with a power supply input of
12 V to check that the output voltage was under the 5 V maximum that could be read by the myRIO.
This was implemented using a 10 k Ω and a 1 kΩ resistor in series and then measuring the voltage
across the 1 k Ω resistor. The voltage drop across any series resistor can be calculated from the
voltage divider equation, Equation 7.6.2, [104]:
Rn
Vn = Vsupply (7.6.2)
Rtotal
With Vsupply = 12 V and Rtotal = 11 k Ω, the theoretical voltage drop across the 1 k Ω is 1.09̇ V .
The results of the bench top test can be seen in Figure 74. As can be seen the voltmeter reads
1.091 V . The difference between the theoretical value and the measured value is likely due to
both the 5% error of the resistors used in the circuit and the fact that the supply voltage is actually
12 ± 0.05 V .
114 CHAPTER 7. ELECTRONICS AND CONTROL
(a) Power supply controlled input (b) Output of voltage divider circuit
Once confident that the voltage divider circuit was safe to use with the myRIO, the complete
voltage divider circuit could be assembled and integrated. The operation of the circuit and the GUI
of the NI LabVIEW software is relatively simple. A warning light is turned on when the voltage drops
below 11.5 V , and then a second warning light and pop up box are activated when the low voltage
disconnect unit breaks the circuit. This operation can be seen in Figure 75.
(a) First LED lights up when the voltage (b) Second LED lights up when the low volt-
across the battery drops below 11.5 V age disconnect cuts the circuit
The script that runs this programme can be seen in Figure 76. The script was first tested with
just the voltage divider circuit and the power supply to check that there were no irregularities. Once
this step was complete, the low voltage disconnect unit was connected up in series successfully.
7.6. ANCILLARY BATTERY SYSTEM 115
Figure 76: Block diargram script that monitors the battery voltage via the myRIO analogue input
This script allows for the voltage of one cell to be monitored. In order to measure the individual
battery voltages when they are in series with one common ground; one voltage divider circuit must
measure the full 24 V, the other then measures one of the individual batteries. The second battery
voltage can then be found by subtracting the individual voltage from the series connected voltage.
At the time of report hand-in, the voltage monitoring circuit was fully implemented on the lo-
comotive without the low voltage disconnects. It is expected that these will be added before the
IMechE competition date.
Current Sensing
The measurement of current flow out of each of the batteries is vital in order to ensure a long
lifespan of the lead acid batteries. As has previously been discussed, the lifelong performance
of this type of battery deteriorates rapidly when the charge of the battery reduces below 30%.
Therefore it is essential to be able to estimate the state of charge (SOC) of both batteries at all
times. To do this, the current out of each battery will be measured by the Hall effect sensor. This
then sends an output voltage signal that is amplified by an operational amplifier and then read by
the myRIO analogue input.
Using the LabVIEW visual interface block diagram code it is possible to read this voltage and
convert it to the corresponding current value (conversion process detailed later in this section). The
charge consumed can then be found by integrating the current with respect to time, Equation 7.6.3
[105]:
Z t
Q= I dt (7.6.3)
0
Where Q is the electical charge, measured in ampere-seconds (As), I is the discharge current
in amperes (A) and t is the time of discharge in seconds (s). The SOC of the battery can then
simply found by multiplying Q by 3600 to get the electrical charge in units of ampere-hours (Ah)
and subtracting this value from the capacity value of the battery, 55 Ah. The theoretical, and yet to
be finalised, current sensing circuit can be seen in Figure 77.
116 CHAPTER 7. ELECTRONICS AND CONTROL
Figure 77: The theoretical current sensor circuit; note that it does not include the wire that carries
the current being measured. The current sensor is represented by the symbol ’X’
The current sensor itself and the test circuit can be seen in Figure 78. The dot on the current
sensor signifies the side through which current flow is positive. This ensures the safe, and correct
operation of the sensor.
(a) Close-up of the current sensor being (b) Current sensor circuit measuring the
used as part of the test circuit current through a contactor switch
The Hall effect current sensor gives a voltage output which much be converted to the corre-
sponding current value. In order to calibrate the sensor, the test circuit shown in Figure 78 was
used to measure the voltage output of the sensor when measuring the current through two contac-
tor switches; the current was measured using an ammeter. The sensor and operational amplifier
were powered using a 1.5 V battery and a bench-top power supply respectively. The results of this
experiment can be seen in Figure 79. Note that due to the usage limits of the contactor switches,
the current range was limited to between 1 and 1.8 A when connected in series.
7.6. ANCILLARY BATTERY SYSTEM 117
Figure 79: The relationship between the measured current and the voltage output of the Hall effect
current sensor
It can be seen that there is a linear relationship which is expected up to 100 A. The equation of
the trendline in Figure 79, Equation 7.6.4, can now be used to convert the sensor output voltage to
the measured current value. It is true that due to the small current range tested, this equation may
not be a true representation of the real conversion factor at higher current values.
Vout + 0.0231
Imeasured = (7.6.4)
0.2103
Due to uncertainty about the current sensor conversion factor at higher currents and also the
final circuit that will be used, it was not included in the locomotive electronic circuit for the pre-report
hand-in testing. Once an opportunity arises to measure higher currents (up to 20 A) the final circuit
can then be added with the security that it will provide the correct readings, refer to Section 14.
Figure 80: Final block diagram of the integrated battery monitoring system
Figure 81: Final front panel of the integrated battery monitoring system
It can be seen that as well as the safety cut-off of the low voltage disconnect. There is also now
an indicator and on-screen pop up for when the estimated SOC of the battery reaches 30% of its
maximum. One major limitation of this setup is that each time the programme is stopped, the data
resets itself, therefore it will be necessary to log the last known SOC of the battery and then use
this as an input the next time the batteries are used. It is hoped to resolve this problem before the
competition hand in.
supercapacitor terminals (see Section 5.6.5); an ABS 3D printed terminal and cable cover is used.
Though in principle the approach taken is similar, complications arose due to the lack of obvious
attachment points for the cover, and the far more dimensionally complicated structure required
to deal with this. In order to prevent the cover coming loose upon locomotive vibration, a tightly
constrained interference fit is implemented between the battery handles and the handle location
groove on the top surface of the cover, which can be seen in Figure 82a. The interference fit can
be seen in practice in Figures 82b and 82c.
As with the supercapacitor cap in Section 5.6.5 the cable approach direction is standardised
for each terminal so as to grant the ability to design in improved protective cover. The cable can
be seen entering the cover in the bottom left of Figure 82b, in which it can also be seen that due
to the standardised approach very little clearance is required, reducing the risks mentioned above.
In order to ensure stability, even when subject to compressive forces a number of stabilising and
strengthening struts are employed, increasing contact with the battery surface, as can be seen in
Figures 82b and 82c.
(a) Battery cover CAD model (b) Isometric view of terminal cover (c) Side view of battery covers
7.7.1 Testing
The main contribution from the Electronics team with regards to the supercapacitor energy recovery
system was to help build and test the simple charge/discharge circuit, provide assistance to modify
the circuit after initial testing and then implement the final circuit on the locomotive.
The initial supercapacitor charge/discharge circuit implemented at the start of testing can be
found in Figure 83. It can be seen that this circuit includes two automated switches that control
when the supercapacitor is charged and discharged.
120 CHAPTER 7. ELECTRONICS AND CONTROL
Through gradual changes in the circuit and continuous testing, this circuit was then optimised to
include a soft charge resistor by-pass switch and switches to discharge and to avoid supercapacitor
overcharge. In the final week of testing, the circuit was also integrated with the motor control circuit
before the full circuit build on the locomotive. All of the switches were automatically controlled by
the myRIO and for testing an additional manual discharge switch was added to the final circuit. In
order to automate this circuit, supercapacitor voltage monitoring was implemented using the same
voltage divider circuit as detailed in Section 7.6.4. The switches were then controlled as follows:
These values are still to be optimised in order to improve the performance of the locomotive.
The front panel and block diagram for the final test circuit can be seen in Figures 84 and 85.
The final switch circuitry used on the locomotive can be seen in Figure 86.
Figure 85: The block diagram that enables the myRIO to be able to control the supercapacitor
circuit
Figure 89: Pin allocation for connecting each subsystem to the myRIO
7.8. SUBSYSTEM INTEGRATION 123
The mounting of components and routing of cable throughout the system proved a challenge
but suitable measures were put in place to minimise the risk of component damage. Trunking was
utilised to protect thinner gauge cable from damage through rubbing on shelves or environmental
exposure. ABS boxes were used to mount and protect constructed PCBs for the voltage, current,
temperature and Hall effect sensors alongside ancillary motor control circuitry and the motor con-
troller. Quick connectors were used to aid modular component removal including molex connectors
and crimps. Subsystems were grouped together to minimise cable runs, primarily the thinner gauge
control wire between the myRIO and subsystems on the rear shelves (the left of Figure 90). The
myRIO was at the rear of the train to reduce the distance to the wireless control device. The batter-
ies, brake system and switch mode power supply was located at the front of the train. Components
were distributed to keep the centre of mass as low and even across the bogies. Components
were secured with locking bolts or Velcro depending on the item weight and vibration isolation was
achieved with foam and rubber isolating feet. The generator and resistors were grouped in the
centre where more intensive cooling can be utilised.
However, the fact that bogies are almost universally used in full-size modern railways does
not necessarily dictate the design decision for a prototype miniature locomotive design, as proven
by previous IMechE Railway Challenge designs, two of which are shown in Figure 91. However,
the fact that bogies are so universal is useful when making other decisions, as there is literature
and previous designs available for locomotive design on full-size trains which does not exist for
miniature locomotives. The decision to use bogies has several other reasons, explained in more
depth below.
The first advantage is in turning; a locomotive design with bogies can handle much tighter
corners since the axles can be fairly close to each other. Initial estimates for the length of the
locomotive were over 2.5 m, which would lead to a concerning distance between axles. Though
there are ways to allow lateral movement of axles, such as the Gölsdorf axle or the Klien-Lindner
axle [110], these methods were used on steam engines and are antiquated, and are mechanically
more complex than bogies.
Secondly, a locomotive design with bogies enables better group workflow. Without bogies,
all decisions regarding wheelsets, suspension, brakes, motors, and drivetrain would have had to
be made with more compromises. Using bogies assures that an ideal environment for all these
components is met.
125
126 CHAPTER 8. BOGIE DESIGN
Lastly, this design choice was made with future years in mind. If the locomotive frame needs
to be made wider or slimmer, longer or shorter, the bogies can remain unchanged. If the compo-
nents on the frame are heavier or lighter, only the suspension needs to be changed. The modular
approach of using bogies will hopefully prove extremely beneficial to future teams.
The downside of this choice is quite simply a more complex overall design, which requires more
parts and has more potential sources of failure.
• Gauge: 10¼ inches (between rail inner faces). May widen to 10½ on curves.
• Minimum horizontal curvature radius: 20 m.
Using this data, a Python script was written to geometrically determine the maximum allowable
distance between the axles in each bogie. The exact explanation will not be given here; however,
the Python script can be found in its entirety in Appendix J The distance between bogies is of
no concern here since each bogie can swivel independently. There are two factors which may
become problematic depending on the curvature radius and bogie size. The first is that the tread
of the wheel must stay on the track at all times; the second is that the wheel flange should not be
rubbing against the inside of the rail.
An initial estimate of 800 mm was made for the total length of the bogie. Since the axles are
fully contained within the bogie, a good first estimate for the distance between axles is 500 mm.
A graphic of this arrangement can be seen on the following page. The view is as seen from
above, at the level of the rail. The two arcs are the rail inner faces, with the centerline between
the two as a dotted line. The flange is represented by a rectangle, as this is the shape of the
cross-section of the flange at the height of the rail. The tread is represented by a line, as in theory
the contact area between a cylinder and a plane (i.e. the wheel tread and the rail) is a straight line.
Furthermore, it should be noted that the wheels are “pressed” up against the outer rail, meaning
there is contact between the flange and the rail. In practice, this should be avoided, but it makes
spotting whether the setup is allowable easier on the inner rail.
In Figure 92, it can be seen that there is a gap between the flange and the rail on the inner rail,
meaning that the setup described above is feasible. Using this geometric method, the program can
be made to run multiple times to determine the maximum allowable distance between axles. The
result of this evaluation is that the maximum allowable space between axles in a bogie is 1.78 m.
This is much larger than the intial concepts, and a bogie of these dimensions will not be used.
8.2. BOGIE FRAME 127
Figure 92: Graphical analysis of wheels and flanges in a 20 m bend radius turn
Therefore, it can be concluded that this particular geometric requirement will not pose a problem to
the bogie design.
These calculations have all assumed a gauge of 10¼"; if the track widens to 10½", the maximum
allowable distance between axles is 5.46 m.
Vertical loading on the central beam occurs during all aspects of operation. During normal opera-
tion, the weight of the locomotive rests on the central bogie beam, leading to downwards vertical
loading, and during jacking or lifting the weight of the bogie is suspended from this beam, leading
to upwards vertical loading.
A rough estimate leads to each bogie weighing roughly 90 kg and the total weight above the
bogies being around 230 kg. This means each bogie carries a load of 115 kg. In addition, this
assumes perfectly balanced weight distribution, which will prove very difficult; in reality, one bogie
will likely need to bear a higher load. For the sake of this calculation, a value of 150 kg can be
used. Since the frame is symmetrical, if this higher value is withstood by the connectors, the lower
128 CHAPTER 8. BOGIE DESIGN
WL
BM = (8.2.1)
8
In this case, the load is 150 kg multiplied by 9.8 m/s2 , and L is 0.3562 m. Therefore, the maxi-
mum bending moment is 65.5 Nm. As previously mentioned, each screw can withstand a bending
moment of 80 Nm, so four screws can withstand a bending moment of 320 Nm, giving a factor of
safety of 4.9.
The horizontal loading on the central beam is due to either maximum acceleration or maximum
braking. The maximum deceleration from braking is specified by the IMechE as 1.3 m/s2 [15]; the
maximum acceleration envisaged is 2.0 m/s2 . Therefore, only the maximum acceleration will be
investigated since the frame setup is symmetrical.
X
Ftrailing max = 600 × 10 sin 2.86 + 60 × 2.0 = 1499.37...N ≈ 1500 N. (8.2.2)
Equation 8.2.2 shows the sum of the forces due to both acceleration and against gravity when
going up an incline. Again, the bending moment equation is as follows [37]:
WL
BM = (8.2.3)
8
Using Equation 8.2.3 and a length of 0.3562 m, the maximum bending moment on the beam is
66.8 Nm. The maximum allowable bending moment, using the already existing connection screws
from the vertical loading, again is 320 Nm since four 80 Nm screws are used. This results in a
factor of safety of 4.8.
The callipers are attached to the end beams, meaning there is a reaction force going through these
beams during braking.
X
Ftrailing max = 600 × 10 sin 2.86 + 600 × 1.3 = 780 N (8.2.4)
This braking force is shared by four callipers, meaning a force of roughly 200 N is applied to each
beam. The length of these beams is also 0.3562 m, leading to a maximum bending moment of
under 9 Nm. Since this force is applied in the horizontal direction and the bending moment is
low, gussets can be used since they are rated in this direction. Each gusset is rated to 60 Nm in
compression and 160 Nm in tension, and since two are used, even during compression there is a
factor of safety of 13.5.
130 CHAPTER 8. BOGIE DESIGN
1.3Mg
dynamic factor of 1.3, as specified in the technical guidance document [15], K = , where M
8
is the sprung mass of the locomotive, and g is the acceleration due to gravity (9.81m/s2 ).
Due to axle symmetry, FR1 = FR2 = K . The bending moment can be calculated using Macaulay’s
method by taking moments about the cut face as displayed in Figure 94b.
Figure 94: A preliminary wheelset setup that can be used to determine requirements
Assuming the cross-sectional area of the axle stays symmetrical about both axes, the following
expression for the bending moment (BM) can be used [111]:
d 2ν
BM = −EI (8.4.2)
dx 2
Substituting this (Equation 8.4.2) into the previous expression (Equation 8.4.1) and integrating
twice, before evaluating the constants of integration based on no deflection at points of reactions
(ν = 0 at x = 0.1m and ν = 0 at x = 0.367m) leads to Equation 8.4.3. This is also plotted in Fig-
ure 95.
EIν h i
= −0.2657 [x − 0.05]3 + [x − 0.417]3 − [x − 0.1]3 − [x − 0.367]3 − 0.04755x + 0.00463
M
(8.4.3)
132 CHAPTER 8. BOGIE DESIGN
Figure 95: Deformation of the axle and outlining the points of maximum and minimum deflection
Using
νthe output function of the Python script used to plot Figure 95, the corresponding value to
EIν maximum
can be determined through finding the difference in deflection between the centre
M νminimum
point of the axle and one of the end points. This is equal to 1.94 × 10−3 . From this, an expression
for the maximum change in deflection of the axle under loading can be calculated as a function of
the Young’s Modulus of the material, sprung mass of the locomotive, and axle diameter. Evaluating
this and inputting known expressions [112] yields the result in Equation 8.4.6. For simplification,
the sprung mass can be estimated to be 750 kg, and a plot that shows how deflection varies with
Young’s Modulus and axle diameter can be constructed, as shown in Figure 96.
νmaximum
EIν
= 1.94 × 10−3 (8.4.4)
M νminimum
πr 4 πd 4
I= = (8.4.5)
4 64
Figure 96: Young’s Modulus of the axle as a function of the axle diameter and sprung mass
Ideally, the differences in deflections should remain at a fraction of a millimetre, and so a com-
bination of a small diameter axle and low value of Young’s Modulus should be avoided. A scenario
where one of either the Young’s Modulus or axle diameter being small is sometimes permissible,
provided the other is increased in value to compensate. If this approach is used, it is more cost
effective and more beneficial practically to have a combination of a high stiffness material with a
lower diameter than a large diameter of axle constructed from a low stiffness material [113].
T τ Gθ
= = (8.4.7)
JT r L
πd 4
If the axle is modelled as a solid shaft, the torsional constant, JT , can be evaluated to
[115].
32
From the torsion equation, two conditions must be evaluated to determine the minimum axle
diameter - one involving the acceptable angle of twist, θ, of the axle; and the other involving the
maximum shear stress before failure, τmax . The minimum acceptable diameter to resist torsion
value will be the maximum diameter value obtained through the analysis of the two conditions.
The Lynch motor being used is quoted to produce 48 Nm of torque by the manufacturer [116].
The torque experienced by the axle depends on the gear ratio that is used. Using the 5:1 gear ratio
figure, as calculated in Section 4.6, with a factor of safety of 1.3 applied to the torque experienced,
134 CHAPTER 8. BOGIE DESIGN
the value of torque at the axle is 312 Nm. As the motor is powering two axles, the actual maximum
torque (and the value used in the analysis) will be half this stated value: 156 Nm.
A Python script was created that housed a dictionary of material properties (those required in
solving the torsion Equation 8.4.7 for the two conditions), and the maximum torque and axle length
(0.51m) that would be used in practice. As the sprockets have not been positioned on the axle, the
full axle length was chosen as opposed to the larger of the two sprocket to end of axle distances.
The material properties required have been sourced from matweb.com [117].
There exists no maximum angle of twist value, however, a fraction of a degree seemed to be
appropriate to set as an upper limit. This decision allowed the axle not to be overspecified, but,
more importantly, will not lead to complete catastrophic failure in service.
Once the script was run, it solves to find the minimum diameter requirement for both conditions
(twist and shear) as outlined in Equation 8.4.7 and prints the larger of the two. If the larger of the
two is below the minimum axle diameter, then the minimum axle diameter is printed, as shown in
Figure 97.
These diameters are greater than the minimum, and thus show that torsion is more of a problem
in this application than bending.
Out of the remaining factors, fatigue is probably the biggest concern. Due to the axle having to
be machined, and it being a crucial part of the drivetrain, frequent replacement or inspections of
the part can prove costly and impractical, thus having good fatigue resistance is important.
It is well known that fatigue life can be drastically improved through simple design improve-
ments [121]. Some common methods include the absence of sharp edges (applying fillets), which
reduces stress concentration at edges; continuous, constant cross-section that aims to reduce
stress concentrations; and having a smooth surface finish, which reduces the chance of leaving
behind residual tensile stresses that would impact fatigue life.
2ηP N ηP N
Tractive Force = = (8.5.2)
ωd u
ηP N
Tractive Effort = + µFR (8.5.4)
u
Maximum tractive force that can be applied is limited by the maximum adhesive force between
the wheel and the rail. Any force higher will cause excessive wheel spin, and so power is being
136 CHAPTER 8. BOGIE DESIGN
wasted. Using the theory of contact mechanics [126], the maximum adhesive force between the
rail and the wheel is given by [127]:
Fadh = 2aνE ∗ (8.5.5)
√
...with νtrack related to the track deflection under loading, a = r νtrack , and E ∗ equating to:
−1
1 − p12 1 − p22
∗
E = +
E1 E2
Subscripts ‘1’ and ‘2’ refer to the wheel and the track respectively.
AE2
The deflection of the track can be found through FR = κνtrack , with κ = , where A refers to
L
the cross-sectional area of the track, and L being the distance between fixed points on the track
(sleepers).
As when analysing axle deformation, the reaction force at each wheel should be taken as FR =
1.3Mg
.
8
Combining the equation describing track adhesion (Equation 8.5.5) and the equation describing
traction (Equation 8.5.4), and substituting in the various expressions listed above, before rearrang-
ing for radius, yields:
" #2
(8AE2 )1.5 1 − p12 1 − p22
ηP N 1.3Mgµ
r= · + + (8.5.6)
2 (1.3MgL)1.5 u 8 E1 E2
Using the simple relation of diameter, d, being twice radius, r , the above equation evaluates to:
"√ 1.5 # 2
1 − p12 1 − p22
2 8AE2 ηP N 1.3Mgµ
d= + + (8.5.7)
2 1.3MgL u 8 E1 E2
Typical values for each parameter can be seen as those displayed in the table below (quantities
can be taken to be assumed, measured, calculated, or taken from matweb.com):
Table 20: Typical values for each parameter that appear in Equation 8.5.7.
Equation 8.5.7 can now be plotted using values outlined in the above table (Figure 98).
8.5. WHEEL SIZE 137
Figure 98: Required minimum wheel diameter to eliminate wheel spin for different masses
Figure 99: Required minimum wheel diameter to eliminate wheel spin for different masses and
differing power outputs
Unfortunately, for a power output of 16 kW, the locomotive would have to incorporate a large
diameter wheel at the target locomotive mass of 750 Kg. This is undesirable, as it means large
amounts of energy need to be expended to move the locomotive, compromising the energy recov-
ery and energy efficiency challenge. Fortunately, the motor can be modified to reduce the power
output, or simultaneously drive two axles, halving the power transmitted to each axle. This means
it would be relevant to see what the ideal wheel diameter would be as a function of both locomotive
mass and power output. This is shown in Figure 99.
It can be seen from Figure 99 that it would be desirable to avoid the scenario (from a traction
138 CHAPTER 8. BOGIE DESIGN
point of view) of a light locomotive with a large amount of power. By reducing the power output to
half its maximum output, a more reasonable value of wheel diameter can be chosen. This is shown
in Figure 100. This can be achieved by the motor driving two wheelsets simultaneously.
The most likely running configuration is having the motor drive each wheelset at 8 kW and a
locomotive mass of around 750 kg. From Figure 100, it would seem the appropriate wheel diameter
would be around 250 mm for this scenario.
Other than the wheel diameter, bore diameter (dictated by axle diameter), and addition of light-
ening holes, the wheel profile has to comply with the drawing specified by the IMechE Railway
Challenge Technical Specification [114]. Thus, detailing, material, and finishing of the wheels are
predetermined. The drawing submitted to the EDMC for manufacture can be viewed in Appendix
Section F as P21-BOG-WHL-001.
Figure 100: Required minimum wheel diameter to eliminate wheel spin for different masses at a
power output of 8 kW
FMAX = 1.3 × (2400g sin (2.48) + 2400 (1.45)) = 5.9 kN (2s.f .) (8.6.1)
For simplicity, and to account for any aerodynamic drag and dynamic friction forces (taken to
be very small compared to component of weight down the slope and the maximum braking force
itself), the maximum longitudinal force can be taken to be 6 kN. As there are four wheelsets, the
maximum force for the number of couplers on a single wheelset is 1.5 kN. This translates to 0.75 kN
on each side of the wheelset.
During jacking, the vertical load on the couplers on each wheelset is simply the weight of the
wheelset itself. The regulations state the maximum wheelset load is 500 kg [123], and so the
wheelset weight can be taken as near enough 5 kN. With a 50:50 distribution on each side of a
wheelset, the vertical load on each side of the wheelset can be taken to be 2.5 kN.
140 CHAPTER 8. BOGIE DESIGN
Figure 101: Highest specification linear motion guide that would fit in the design space [129]
The largest linear motion guide that would fit in the design space is shown in Figure 101. As-
suming four wheelsets that are somewhat symmetrical, the maximum moment due to braking can
be estimated to be the braking force supplied by one wheelset multiplied by half the track gauge.
This equates to roughly 110 Nm. This is almost 20 times the maximum moment the linear motion
guide is rated to, and so would be unacceptable to use this as a solution. Using a large number
in series would be impractical, and any misalignment with one linear motion guide would cause
problems for all the others. These factors require a custom design needs to be developed.
After consulting with other team members, it was decided a metal plate bolted onto the bogie
frame with a cut-out for a pin constricted to the wheelset to slide into would be a suitable solution.
A visual representation of this system is shown in Figure 102.
Figure 102: Implemented version of the custom designed coupling designed to eliminate longitudi-
nal relative motion
The key features of note include the fixing of the plate to the frame, the slot restricting wheelset
motion so no relative longitudinal motion can occur, and the number of plates per side (four plates
8.6. WHEELSET TO BOGIE FRAME COUPLING 141
per wheelset, two per side) to allow for stress to be distributed over a greater number of pins. The
pin on the upper suspension adapter is not strictly necessary - the load can be transferred to the
top bolts on the extruded aluminium. Its primary purpose is to keep the suspension adapters lined
up with each other. The final job before manufacturing the coupling is determining its dimensions,
along with the dimensions of the pin.
The distance between the two reaction forces is 0.035 m, and the distance between R2 and Fb
is 0.079 m. Performing similar analysis as in Section 8.4.1 yields R1 = 1066 N; R2 = 691 N and
M = 8.0625 Nm.
Using the intermediate bending moment equation (not listed), it can be seen the maximum
bending moment magnitude occurs at the location of the upper pin (location of R1 ). The magnitude
of the maximum bending moment at this location (y = 0.035 m is 16.2 Nm). From this, the di-
mensions of the plate cross-section can be determined using the symmetrical cross-section stress
equation (Equation 8.6.2 [111]).
Mx,max zmax
σy,max = (8.6.2)
Ixx
tb3
Ixx = (8.6.3)
12
The thickness, t, of the coupler plate (Figure 103) should be something that can be readily pur-
chased. If it is assumed the plate would be made from aluminium to be consistent with surrounding
metals, and the worst case yield stress is used (σy,max = 15 MPa [130], even though in reality
a quality grade of aluminium will exhibit a yield strength of greater than 200 MPa [131]), dimen-
sion b can be solved for (Equation 8.6.4) through combining and rearranging Equations 8.6.2 and
142 CHAPTER 8. BOGIE DESIGN
1/2
6Mx,max
b= (8.6.4)
tσy,max
Using a plate thickness of 10 mm, the minimum length b works out to be 0.02 m. The analysis
has been simplified, however, and does not account for the weakening of the plate due to having
a slot cut out. As such, a safety factor of 2 will be applied on the the plate cross-section length, b,
resulting in it being 40 mm. If an alternative material is used, the thickness of the plate that would
be required can be found through Equation 8.6.5 [132].
s
YSAl
tmat ≥ tAl (8.6.5)
YSmat
As the yield strengths of strengthened aluminium and mild steel are very similar, the materials
can be interchanged based on cost grounds with negligible effect on dimensions of the cross-
section.
The next step is to dimension the pins. The load due to braking has already been calculated for
both pins, and it is assumed that this is evenly distributed across the cross-section thickness, t.
The other loading condition is for when the locomotive is jacked up. This will only affect the
bottom pin as it is the pin connected to the wheelset which will effectively be suspended from it. It
was previously stated that the wheelset would have a maximum weight of around 5000 N. As there
exist four pins per wheelset, this load can be taken to be 1250 N per pin.
The above discussed loading conditions can be visualised over the pin as shown in Figures 104a
and 104b.
There is now the need to calculate the direct stress due to bending, and the shear stress before
resolving into a von Mises stress distribution across the cross-section.
To calculate the maximum bending moment experienced around each axis, the principle of
minimum total potential energy will be used [133]. This states that equilibrium is reached when the
variation of the total potential energy of the system is equal to 0, as expressed mathematically in
8.6. WHEELSET TO BOGIE FRAME COUPLING 143
Equation 8.6.6.
The stored potential energy, PEst , for a bending scenario can be found to be equivalent to [134]:
t 2
d 2ν
Z
1
PEst = EI dx (8.6.7)
2 0 dx 2
The potential energy due to external loading, PEext , is equal to the negative of the work done
[135], WD, in the system by external forces. In the case of the top pin, this evaluates to Equa-
tion 8.6.8.
Z t
R1
PEext = −WD = − · νdx (8.6.8)
0 t
Thus, the total stored potential energy in the system (for the top pin) becomes:
Z t 00 2 Z t
1 R1
PEtotal = EI ν dx − · νdx (8.6.9)
2 0 0 t
For equilibrium, the variation in total potential energy (δPEtotal ) must equal 0. When applied, it
results in Equation 8.6.10.
Z t Z t
1 00 00 R1
δPEtotal = EI 2ν δν dx − δνdx (8.6.10)
2 0 0 t
00
δν has no meaningful use, and so the first part of Equation 8.6.10 must be continually inte-
grated by parts until all integrands appear with δν. Performing the integral twice yields the result in
Equation 8.6.11, which can be rearranged into the form shown in Equation 8.6.12.
" # Z
h 00 0 it h 000 it Z t 0000 t
R1
δPEtotal = EI ν δν − ν δν + ν δνdx − δνdx = 0 (8.6.11)
0 0 0 0 t
h
00 0 t
i h 000 it Z t 0000 R1
δPEtotal = EI ν δν − ν δν + EIν − δνdx = 0 (8.6.12)
0 0 0 t
From Equation 8.6.12, the natural boundary conditions can be found through evaluating the
non-integral terms. This equation also yields the governing equation that can be further worked on
to find the maximum bending moment.
As δν is arbitrary [136], for the entire equation to equal 0, each term must equal zero, and,
more importantly, the integrand itself must equal 0. This results in the following five conditions or
expressions.
00 0
ν (t) δν (t) = 0 (8.6.13)
00 0
ν (0) δν (0) = 0 (8.6.14)
000
ν (t) δν (t) = 0 (8.6.15)
000
ν (0) δν (0) = 0 (8.6.16)
d 4ν R1
EI − =0 (8.6.17)
dx 4 t
144 CHAPTER 8. BOGIE DESIGN
0
Looking at the geometry in Figure 104a, it can be said that when x = 0, δν = 0 and δν = 0
0
(fixed wall boundary conditions). When x = t, δν 6= 0 and δν 6= 0 as it is a free end. Implementing
these last two conditions into Equations 8.6.13 and 8.6.15 means:
00
ν (t) = 0 (8.6.18)
000
ν (t) = 0 (8.6.19)
These are known as the natural boundary conditions, and become useful when evaluating the
constants of integration when integrating Equation 8.6.17 twice to obtain an expression for the
d 2ν
bending moment (since the magnitude of bending moment equates to EI 2 ). Physically, Equa-
dx
tion 8.6.18 indicates the bending moment is 0 at the free end, and Equation 8.6.19 indicates the
shear force is 0 at the free end.
Integrating Equation 8.6.17 twice, inputting the natural boundary conditions to evaluate the
constants of integration, and using Equation 8.4.2 yields an expression (Equation 8.6.20) for the
bending moment across the pin. This can be plotted (Figure 105) to find the location and magnitude
of the maximum bending moment.
R1 t R1 x 2
|BM| = + − R1 x (8.6.20)
2 2t
Figure 105: Bending moment variation about the y-axis for the top pin
The maximum bending moment occurs at the root, as expected, and has a magnitude of
5.33 Nm for the top pin. As there is no loading parallel to the y-axis for this pin, the maximum
bending moment about the z-axis is 0. This means the stress distribution over the cross-section of
the top pin at the root of the pin can be described by Equation 8.6.22. This is the simplified version
of Equation 8.6.21 [137].
Mz Iy − My Iyz y + My Iz − Mz Iyz z
σx = 2
(8.6.21)
Iy Iz − Iyz
5.33z
σx = (8.6.22)
I
8.6. WHEELSET TO BOGIE FRAME COUPLING 145
πd 4
Both second moments of area, Iy and Iz , for the pin are equal to , and is denoted I. Due to
64
the symmetrical cross-section, Iyz = 0 [137].
This analysis can be performed for the bottom pin. When performed, the maximum bending
moment occurs at the root of the pin, and the stress distribution of the bottom pin across its cross
section is given by Equation 8.6.23.
6.25y − 1.875z
σx = (8.6.23)
I
The final components of stress the pin can experience is shear stresses on the x-face in the
direction of y, τxy , and z, τxz . Knowing the relation described through Equation 8.6.24 [138], and
applying it to Equation 8.6.21 for the symmetric pin cross-section yields equations for shear stress
distribution across the cross-section (Equations 8.6.25 and 8.6.26) [139].
Z
1 dσx
τ= dA (8.6.24)
thickness A dx
16F d 2 − 4y 2
τxy = = τyx (8.6.25)
3πd 4
16F d 2 − 4z 2
τxz = = τzx (8.6.26)
3πd 4
The equations describing stresses experienced (Equations 8.6.22, 8.6.23, 8.6.25, and 8.6.26)
can be used to calculate the relevant stress components, which can then be combined to calculate
the von Mises stress distribution across the cross-section of the pins. The von Mises stress equa-
tion for the pin loading described through Figures 104a and 104b is stated in Equation 8.6.27 [140].
q
σvon = σx2 + 3 τxy
2 + τ2
xz (8.6.27)
The von Mises stress distribution can be plotted for different diameters of pins which would help
determine the ideal pin dimension. Assuming bolt construction is of A2-70 Stainless Steel, it would
be desirable to keep the maximum stress experienced well below the 0.2% proof stress value of
450 MPa [141].
The plots below (Figures 106a and 106b) show stress distribution for a pin under worst case
loading conditions. Stress analysis for an ideal pin diameter of 8 mm is shown; The Python script,
however, allows a pin of any diameter to be investigated. Figures 106a shows the stress distribution
for the top pin, and Figure 106b shows the stress distributions for the bottom pin.
146 CHAPTER 8. BOGIE DESIGN
(a) von Mises stress distribution across the top pin (b) von Mises stress distribution across the bottom
when it is assumed to have a 8 mm diameter pin when it is assumed to have a 8 mm diameter
Figure 106: von Mises stress distribution on the top and bottom pins for an 8mm pin diameter
8.7 Suspension MD
Double Suspension
The concept of double suspension is to add an additional layer of suspension between the final
sprung mass and the rail-wheel interface. If used correctly, this can lead to improvements in sus-
pension characteristics and a lessened effect of rail or wheel irregularities on ride comfort. This
might seem to be an obvious choice if ride comfort is a priority (which in this case it is, as the
IMechE Railway Challenge includes a ride comfort challenge); however, like any additional feature,
there are also costs and trade-offs that must be considered.
The trade-offs, in this case, are cost, weight, space, and complexity. In the case of this specific
train design, cost is not a crucial factor, and using a second layer of suspension would most likely
be financially feasible. Weight is also not a concern, as the design is predicted to be significantly
under the IMechE weight limit of 2,000 kg. Space, however, is a concern due to the large size of
the motors which have to be mounted on the bogie, in a fairly central position, which is the position
a potential secondary suspension would also be vying for. Air suspension systems (explained in
more detail in the following section), which are the most commonly used type of suspension system,
take up quite a bit of space, and the overall size and shape of the bogie would likely have to be
8.7. SUSPENSION 147
altered to fit both the motor and an air suspension system. The final trade-off is actually the most
important: complexity. During the entire design process, a concerted effort has been made to keep
the design as simple as possible. This is especially important as this is the first year the University
of Southampton will be competing in the IMechE Challenge. A more complex design introduces
more potential points of failure, which should be reduced at all cost.
The benefits of double suspension are clear; whether these benefits outweigh the trade-offs will
become apparent after some analysis and experimentation.
Leaf springs used to be a common suspension system; however, their use has declined significantly
over the last century. Although leaf springs can provide damping due to inter-leaf friction, it is
difficult to calculate this damping and varies significantly depending on lubrication or contamination
of the surfaces. This means that damping cannot be as targeted as a separate damper, and leaf
springs are not as simple as coil springs.
Air Springs
Air springs are commonly used in modern passenger wagons, though generally only as a sec-
ondary suspension.
Coil Springs
Coil springs have been and still are in wide usage for many suspension applications, due to their
ease of use, production, and maintenance. Furthermore, they are lightweight and small. Some
disadvantages include the inability to vary spring characteristics in operation, as well as the fact
that damping is insignificant on coil springs and thus separate damping must be used.
Rubber Springs
Rubber springs work by elastically deforming a block of rubber. One benefit over coil springs is
that rubber springs exhibit better damping properties. However, rubber and other plastics are not
as durable as metal coil springs.
Active suspension
Active suspension does not use the compression of a material to provide suspension. Instead,
actuators are used to determine the position of each individual wheel, giving very precise and
advanced suspension features when done correctly. However, the amount of electronics required
and complexity of the entire system do not make this a viable option, even in full-size trains.
Pst
fst = (8.7.1)
c
where fst is the displacement, Pst is the normal force, and c is the stiffness.
148 CHAPTER 8. BOGIE DESIGN
Equation 8.7.2 relates Hooke’s law to the body bounce of the sprung mass.
c g
ω2 = = (8.7.2)
M fst
The body bounce frequency ranges anywhere from 0.9 Hz to 4 Hz for a fully loaded freight
wagon [107]. Unfortunately, there are no indications for locomotives, and since no further specifics
are given, a range of values will have to be assumed for the moment.
A further complication was that during the early stage in the design process, the sprung mass
of the locomotive was unknown. The competition requires the total mass to be under 2000 kg,
though initial estimates were closer to 1000 kg. Having changed from a fully electric to a petrol
electric traction system, estimates were lowered even further. This means that the sprung mass
may be anywhere between 400 kg and 800 kg. Using these rough mass estimates, as well as
the vague frequencies of body oscillation between 0.9 Hz and 4 Hz, the stiffness c is anywhere
between 325 kN/m and 12,800 kN/m.
Furthermore, these basic calculations hinge on the body bounce frequencies given in Iwnicki
et. al., which are not only vague but are also not necessarily equivalent on 10¼ gauge track.
In the Handbook of Vehicle Dynamics, a model bogie built for testing purposes (with roughly
similar dimensions to those being investigated) used a vertical stiffness of 4,730 kN/m; however,
the body mass was only 33 kg [107].
Experimental Setup
Testing was performed at Eastleigh Lakeside Railway using the “Eurostar” locomotive pulling sev-
eral passenger carriages. Two accelerometers were used, of type Brüel & Kjær 4514-001, kindly
lent to the team by the ISVR. During each test run, both accelerometer inputs were measured.
These were fixed to different parts of the train, with one unsprung and one sprung accelerometer
measurement for each run. The first accelerometer was fixed to the pivot which the axle is fixed
to, which effectively acts as an axlebox, on a passenger carriage. The second accelerometer was
placed on various sprung parts of the train:
While ideally it would have been preferable to mount the unsprung accelerometer to the lo-
comotive when performing tests on the locomotive bogie frame, this was not possible due to the
design of the locomotive bogies, which do not provide enough space on the axlebox to mount an
accelerometer. In theory, there should be no difference in the vibrations experienced by the pas-
senger bogie axle and the locomotive bogie axle, since the track and wheel size are the same.
However, one issue to consider is that in the time domain, the locomotive and passenger carriage
pass the same section of track with a slight delay, so there should be a slight, but noticeable time
offset when passing over a rail joint.
Data Analysis
Several test runs were made, all achieving similar results. The calculations and analysis will be
performed for one such test run; the other test runs can be found in Appendix I.
The graphs of the data collected from one such test run can be seen in Figure 107. The two blue
graphs, i.e. the first and third graphs from the top, are the measured accelerations, on the unsprung
and on the sprung mass respectively. The second and fourth figures are the Fourier transform of
these two graphs, which changes them from the time domain to the frequency domain. In the fourth
graph, some trends are already apparent, especially in the low frequency range.
By dividing the sprung frequency plot by the unsprung frequency plot, the fifth and final figure
is generated. This isolates the effect of the suspension, and shows which frequencies are damped
most by the suspension. The main vibrations that are being damped by the springs are in the low
frequency domain, around 15 Hz. From this, the required stiffness can be found using the following
equation [142]: r
1 k
fn = (8.7.3)
2π m
where fn is the natural frequency, k is spring stiffness and m is sprung mass.
15 km/h. However, if in future the locomotive’s primary power source is replaced by a hydrogen
fuel cell or some other silent technology, noise reduction in the wheelset interactions will need to
be considered.
P21-BRK-001.
The two outputs from the FEA show the maximum stress experienced under the loading, along
with the safety factor. In reality, the locomotive will not be 2 tonnes in mass (most likely around
500-750 kg), and so a safety factor greater than 1 would show a design suitable for the operating
environment. The safety factor calculated through FEA shows it to be 1.14. This is sufficient to use
this design.
Figure 108: FEA output showing the maximum possible stress under braking a locomotive of mass
2 tonnes pulling a 400 kg trailing load
Figure 109: FEA output showing the safety factor under braking a locomotive of mass 2 tonnes
pulling a 400 kg trailing load
8.9. FINAL DESIGN 153
8.8.2 Callipers
Airheart MB1 mechanical brake callipers were selected; these come with an adapter on the top
side of the calliper with holes for bolts. Ideally, it would have been possible to have aluminium
frame directly above these holes. However, this was not possible since the height of the calliper
did not match up with the height of the frame, so an adapter was designed to mount the callipers to
the frame. The design is a simple L-bracket made from 2 mm sheet steel. This was validated using
FEA under maximum braking conditions of 800 N on each calliper (see Table 15 in Section 6.9);
the results are shown in Figure 110b. The safety factor under these worst-case conditions is 1.7.
8.8.3 Cylinder
The actuating cylinder needs to be mounted within reach of the callipers, so as to move the lever
situated on the callipers. The cylinder has to be mounted in such a way that it actuates tangentially
to the brake disc, as this is the direction the lever needs to be applied to apply the brakes. The
cylinder has two dowels on either side which can be used for mounting. One issue that came up
was that the opening of the cylinder, which the tubing connects to, is directly behind one of the
dowels; this makes cylinder placement slightly more involved because it is not possible to simply
put two beams on either side of the cylinder.
The solution devised is to use short sections of aluminium beam, which have holes in the centre.
So as to reduce them
8.9.1 Wheelset
The final wheelset-suspension assembly is shown in Figure 112.
A fundamental part of the locomotive is the frame or chassis, which holds together the other sub-
systems and links the bogies to the trailing load. The “chassis” for a train was often called a
locomotive frame when referring to older models, particularly during the steam train era. As part of
this project the three main options for frame design were identified and compared:
• Full size frames have traditionally been cast [145], ensuring high strength and creating the
correct shape. This was used extensively when it was necessary for intricate details to be
included to allow the placement of boilers and associated machinery. For the purposes of
this project this is impractical, requiring significant investment in tooling, which makes casting
uneconomic for building a single example. If an intricate design was generated and mass
production was needed, casting the frame could be considered.
• More modern locomotives use welded steel frames to form the base on which to build other
parts of the structure [146]. Although a cheap and practical solution which would be more
than capable of withstanding the loads involved, welded steel frame cannot be easily changed
once manufactured, requiring the design to be 100% complete before any manufacture can
take place. Because the requirements of this project will develop as the project progresses,
this method may not be suitable as it does not allow these changes to be accommodated.
The initial cost of materials for this method is approximately £260, but there would also be
manufacturing costs incurred to weld the frame together.
• A potential solution seen on similar prototype small locomotive designs uses extruded alu-
minium beams with adjustable fixings. The beam provides the strength needed, however the
nature of the fixing mechanisms make designing sufficiently strong structures more compli-
cated.
For the remainder of this report only the latter two options were explored.
9.1 Loading DH
In order to evaluate the different options, the loading associated with the train was calculated to
generate a load situation. The rules stipulate a 600 kg mass (m) trailing load must be pulled, with
track conditions including a 5% gradient [15]. The maximum breaking deceleration is specified as
-1.33 m/s2 (+ 0.15), however the maximum acceleration (a) is predicted to be 2 m/s2 .
X
F = m × (g × sin(θ) + a) (9.1.1)
With a gradient of 5%, the angle of incline (θ) is 2.86°. The gravitational constant (g) is known
to be 9.81 m/s2 . Inputting these values into Equation 9.1.1 gives Equation 9.1.2.
The IMechE specification also requires a 30% safety factor to be included, as shown in Equa-
155
156 CHAPTER 9. FRAME DESIGN
tion 9.1.3.
The trailing load also generates a moment in the vertical plane, due to the connector being
mounted below the rest of the frame. The rules specify the connection bar must sit 10” above the
rail head [15]. Once the height of the bogies and bearing were factored in, the connector would sit
115 mm below the top of the frame. The vertical moment can be calculated giving the torque (T )
applied on the headstock, as shown in Equation 9.1.5.
The bogie connection must also withstand the mass of the locomotive acting vertically through
the supporting beam. Because the mass of the locomotive may increase as systems are added,
and will also be unevenly distributed during travel over inclined sections of track, the connections
should be able to withstand more than half of the mass situated on the frame, to allow for a suitable
safety factor. The locomotive mass is predicted to be 500 kg, with 250 kg situated on the frame,
the bogie beams should therefore be able to withstand 1500 N acting vertically on the centre of the
beam.
The Bosch Rexroth system was eventually selected by the project team due to ease of availabil-
ity, being stocked by RS Components for next day delivery, as well as being able to take advantage
of Siemens 20% RS discount.
Bosch’s range include a variety of thinknesses of beam. By using the graph in Appendix H the
45×45 mm beams were chosen based on the loading conditions. This size of beam also allowed
access to the largest variety of couplers in the range, ensuring that any future requirements could
be met by connection systems.
Most manufacturers provided CAD files of the extrusion profile to aid drawings. Bosch’s profile
was imported and extruded to form the construction beams for CAD modelling.
The final design was reached by the processes outlined above. The main issue with the alu-
minium extrusion is that the connectors are the likely location of failure, being far weaker than the
beams themselves. This makes the connection calculations vital, but it is also important to confirm
the stress experienced by the beams does not exceed the stated yield stress.
A second option of having a welded steel frame was pursued using the dimensions and the
loading outlined in Section 9.1. The viability of the backup was validated using basic FEA modelling,
as shown in Figure 113, which confirmed steel frame was suitable for the project if necessary.
Figure 113: von Mises FEA stress plot for steel frame design
9.3 Layout DH
Initial dimensions of 700 mm wide and 3000 mm long were used when calculating original materials
costs. This was generated from the maximum allowable width (710 mm) as per the IMechE Railway
Challenge Technical Specification [15], incorporating space for protruding bolts, and the length of
the previous Eurostar train being roughly 3 m. The aim was to decrease these figures as more
information became available.
The biggest factor in frame length choice was how much room the bogies would require to
rotate. To ensure there would be no interference under any conditions the diagonal distance of the
bogies was used to ensure the bogies would not touch the headstock beams. The bogies were
also mounted so that they would not touch in the centre, allowing 1035 mm between bogie-frame
connections. The clearance between the headstock beams would therefore have to be a minimum
of 1830 mm. This meant the frame length was finalised with 2 m long longitudinal beams.
Several factors influenced the frame width, with the overall constraint stipulated in the rules be-
ing that the train must be less than 710 mm wide as identified above. This must also include the
shell and any control equipment mounted on the side of the locomotive plus a suitable safety mar-
158 CHAPTER 9. FRAME DESIGN
gin. The minimum width is influenced by the width of the bogies and also the generator, ensuring
enough room to access machinery and allow for cable management.
As will be explained in Section 9.4, where possible beams are laid onto of supporting beams.
This is the case for the beams supporting the generator, as can be seen in Figure 114 .
Figure 114: View of the frame showing the lateral generator support beams fixed on top of the main
longitudinal beams
9.3.1 Superstructure DH
Another requirement of the frame structure is to support the subsystems and integrate them into the
rest of the frame. Initially the project team suggested having metal plates (as seen in Figure 116a)
bolted onto the frame. Three separate plates were used to facilitate easier disassembly, as well as
to isolate the vibrations coming from the generator mounted centrally. As the space requirements
of the subsystems became more apparent, coupled with the need to have efficient cable manage-
ment, it was clear there would not be enough space on the floor alone. A system of racks was then
created, situating the heaviest and least accessed parts (the batteries and supercapacitor) on the
bottom to ensure a lower centre of gravity, with the smaller and more accessible components sitting
on the top levels.
Using a plate metal floor for the generator not only seemed wasteful, but would also provide
a large diaphragm to amplify any vibrations produced. This lead to three beams being chosen
to support the generator. Another course for concern were the two resistors used as part of the
supercapacitor charge and discharge circuits. Not only were they extremely large compared to
other circuit components, but they were also predicted to reach high temperatures during operation,
requiring significant cooling and requiring a stand-off distance from sensitive components such as
the supercapacitor. Because there was significant space, and a larger volume of air for cooling, the
resistors were placed either side of the generator, allowing cooling vents to be specifically cut into
the shell.
During assembly, bolting the lower shelves straight onto the frame was experimented with,
as seen in Figure 115 , but because some of the connectors interfered with gussets, as well as
vibrations causing the unclamped sides to knock against the frame, meant the beams were used
instead. This also gave better access during assembly to the bogies below.
9.3. LAYOUT 159
Because the racks would mostly experience small loads, S12 connection screws as detailed
in Figure 120b were used instead of gussets as will be discussed in Section 9.4. This reduced
costs but required the vertical beams be drilled to allow the screws to be tightened, increasing
machining time and reducing flexibility. This required shelf dimensions and layout to be finalised
before construction, factoring in the heights of components including their connector covers and
dampening inserts.
As the frame and racks moved towards a final design there was a need to integrate the frame
and shell to provide support. During consultation with the shell design team it was decided the side
plates would bolt directly onto the vertical beams of the racks, but that the curving roof would need
its own support to give the desired shape. This meant adding an extra longitudinal beam on top of
the racks, providing connection at the correct height to attach the roof. The height increase was
facilitated through use of a double height (45×90 mm) beam, bolted on using 90×90 mm gussets.
160 CHAPTER 9. FRAME DESIGN
By resting the ends of the top shelf beams on the vertical struts, any downward force produced
by the roof would be transferred directly to the frame, without the need to increase loading on
the connection screws holding the rest of the shelves together. A beneficial side effect of the
longitudinal top beam was that it would act to strengthen the frame against bending, providing
extra reinforcement through compressive strength.
FEA was conducted on the shelf as seen in Figure 117. This used the highest predicted loads
which were present on the battery shelf due each of the two batteries weighing 16.5 kg. By having
5 mm aluminium shelves there was a 100% safety factor, allowing for the assumptions of the FEA
as well as any holes drilled in the plate when attaching equipment, as seen in Figure 118.
Figure 117: von Mises FEA stress plot for 5 mm aluminium shelf design
Figure 118: Contactors being attached to the shelf by the E&C subteam
9.4 Connections DH
As part of the Bosch Rexroth system there are many different connectors available, each cover-
ing specific loading conditions. Some connectors require preparation and cutting of the aluminium
beams in order to be located in place. This would somewhat nullify some of the reasons for choos-
ing the aluminium extrusion system, where the beams would no longer be easily moved. The main
connection method chosen was therefore connection gussets which can be bolted and un-bolted,
allowing movement up and down the beams. Gussets also give good strength relative to other
connection methods and are therefore able to withstand the loads outlines in Section 9.1.
The extracts in Figures 120a and 120b are taken from the Bosch Rexroth catalogue and show
the load ratings for the connectors chosen. As shown in Figure 120a the 45×90 mm gussets
provide the strongest fittings but are rated higher in tension than compression. This meant in many
9.4. CONNECTIONS 161
early designs that gussets were used on both sides of the headstock, causing the beams to be set
back from the end of the frame as shown in Figure 119. This was not desirable due to the extra
length needed and potential conflict situations when integrating with the shell. This was overcome
by butting the headstock beams against the ends of the long side beams. This caused the forces
to be directly transferred onto the beam in one direction, halving the gusset requirement. For the
smaller load-bearing connections the 45×45 mm connectors provide enough stability, are cheaper
and take up less room than the larger gussets. This made them the standard choice for this project.
(a) Gusset connector load ratings [147] (b) Bolted connector load ratings [147]
Figure 121: View of the headstock beam showing the 45×45mm connectors attached between the
longitudinal beams and the headstock
Equation 9.1.5 also shows that the headstock must withstand a torque of 583 Nm in the vertical
162 CHAPTER 9. FRAME DESIGN
direction. By fitting the double height headstock beams directly on the end of the longitudinal side
beams, there is space for a small 45×45 mm connector under each side of the headstock as shown
in Figure 121. These connectors can withstand a force of 3000 N as shown in Figure 120a, which,
when acting with the vertical moment arm of 45 mm, can withstand 135 Nm each as shown in
Equation 9.4.1.
When both small connectors are combined with the 60 Nm torsional strength from each of the
45×90 mm gussets, the combined torsional strength is 390 Nm, showing that together they can
withstand the torque calculated in Equation 9.1.5 and provide a 70% safety factor.
Because the beam has fixed ends attached to the frame, a different bending moment calculation
can be used [37]:
(a) Connection screws in double beam (b) S12 connection screw arrangement
1500 × 0.51
= 95.625 Nm (9.4.3)
8
The S12 connection screws are rated for 80 Nm each as shown in Figure 120b and can with-
stand 1300 N acting at the connection. By using two each side of the bogie beam, they alone are
enough to withstand the bending moments and vertical forces (explained in Section 9.1) acting on
the connection, with the large gussets providing an extra safety factor in addition to their primary
role of providing horizontal strength.
Equation 9.4.3 can also be used when calculating the longitudinal forces acting on the beam.
By using Equation 9.1.6 to give the load acting on the beam, Equation 9.4.4 gives the bending
moment on the bogie beam.
2751 × 0.51
= 175.37 Nm (9.4.4)
8
By using the large gussets identified in Figure 120a there is a large safety factor, as each
gusset can withstand 180 Nm in their weakest direction, and 400 Nm in their strongest, giving a
total strength of 1160 Nm for each beam.
When working with connectors it is important to follow the manufacturers torque recommenda-
tions when assembling the frame. For the gussets, the bolts are required to be tightened to 25 Nm,
if the design strengths are to be used, as shown in Figure 120a. This requires use of a torque
wrench during construction.
9.5 Coupling DH
As described in Section 9.1, the coupling connector must allow the coupling bar to be attached
10" above the rail head and must be able to withstand the coupling loads as calculated in Equa-
tion 9.1.2.
164 CHAPTER 9. FRAME DESIGN
Initial designs used the single story 45×45 mm headstock beam as shown in Figure 124, but
these would have been unable to withstand the torsional moments as described in equation 9.1.5.
To mitigate this, a double story 45×90 mm beam was introduced with small gussets attached
underneath as shown in Figure 121. By mounting them underneath, the gussets are loaded in their
strongest direction, allowing them to withstand to forces required. Using double story beam also
gives two attachment points for the coupling connection, and moves the connection point much
closer to the required drawbar height, therefore reducing the moment arm.
Although connectors which would wrap around the beam were considered, as seen above, bolt-
ing the connections directly to the beam would also be strong enough and be a simpler design. A
potential downside would be interaction between the bolts and the connection bar, but calculations
confirm that at the specified height the connection bar will not interfere with the bolts. FEA was then
conducted on the final design which confirmed the suitability of the material for the forces involved.
Figure 125: Von Mises FEA stress plot for coupler design
9.5.1 Drawbar DH
As well as the requirement for the coupler there was also a need to manufacture a drawbar to
connect to the trailing load. As part of the competition there was a specification and technical
drawing to follow which can be seen in Appendix G, which will make manufacture simple since the
material and bearings have been purchased already.
9.6. JACKING 165
9.6 Jacking DH
In order to move the locomotive into place there is a need for a jacking system to lift the locomotive
clear above the track and move it around. The rules also require jacking points to be identified and
marked.
Initial research was conducted into different methods, specifically to whether separate systems
should be used to jack and move the locomotive, or if a single system would be a better option.
Whatever system is selected, a suitable location will need to be identified which can withstand
the whole mass of the train. Ideally this should be close to the bogie connections, which are high
concentrations of load. By locating the jacking points on the on the main beams there is no need
to add extra strengthening methods and should therefore make the system simpler.
As discussed above, parts of the design were changed to accommodate the extra vertical forces
associated with jacking the locomotive. This initially meant extra beams were added above the
bogies to withstand their weight while hanging beneath the frame, but as discussed in Section 9.4
the final design used connection screws.
A system which would allow both outcomes to be achieved could use trailer jockey wheels,
which provide a load bearing wheel and a way of raising the frame. The jockey wheels would
need to be located a suitable distance from the frame to allow the handle to rotate while raising
or lowering the frame. This requirement lead to the use of a steel box-beam to allow the wheels
to be offset from the frame. The jockey wheel attachment plate was the bolted onto the side of
the beam to reduce the manufacturing requirement. To ensure the wheels did not fall over, two
methods were suggested. The first involved a second beam running lengthways along the side of
the frame, this would transfer any rotational motion to the long bar rather than toppling the wheels.
This method would put less stress on the connection but would take up more room and require
more machining. The second method was to bolt the main steel beam directly onto the bottom of
the frame, the bottom was suggested as an attachment point to avoid the need for gussets, which
have low torsional strength, and also so there would be no interference with the shell on the outside
of the frame. The downside of this was the higher stresses transferred onto the connection with the
frame.
An initial design is shown in Figure 126a showing a single beam which would fit under the frame
next to the headstock beams.
This design then evolved to include two supporting angle brackets, as seen in Figure 126b,
which would prevent the beam rotating when the wheels were aligned at certain angles. Connection
grooves were then added to these supports to allow them to be fixed onto the underside of the
frame. The assembly was then welded together with holes drilled to allow the attachment of the
jockey wheel clamps on the side.
One of the considerations when designing this system was the overall width. Although the aim
was to allow it to fit on the tail lift of a Luton van, there was also a requirement to allow sufficient
room for the handle to rotate. The final design meant reducing the radius of the handle to avoid
impacting the shell while still allowing the system to fit within the constraints.
9.7 FEA MC
In order to validate the final frame design, FEA was conducted. With the geometry confirmed, the
material was set to Aluminium 6063 T6 as indicated by the manufacturer technical specifications
for the Bosch Rexroth frame [147]. The geometry used was the same as indicated in Figure 114
and the FEA set-up is shown in Figure 127a. The frame FEA was carried out to replicate the
locomotive frame under loading from components above it, resting on the bogies and pulling a
trailing load. A central load of 300 N spread across the central three support beams to replicate the
generator was included along with four point loads of total force 400 N at each end of the frame to
simulate the shelves supporting electrical components such as the supercapacitor and batteries. A
force applied to the frame by the trailing load was also included, see below for details. Forces are
indicated by purple arrows and blue arrows indicate the rigid connections between frame pieces
to simulate the gussets. Note that in this simulation, the in-plane internal corners have larger split
curve surface areas to simulate the more highly rated gussets for use in these instances. Green
arrows represent the fixed geometry which was chosen to be the connection between frame and
the bogie-frame bearing.
As mentioned above, a force applied to the rear of the frame by the trailing load was included in
the simulation. This force is the result of the locomotive pulling a 600 kg trailing load up a maximum
incline of 1:20 or 2.86°at the maximum predicted acceleration of 2.0 m/s2 under ideal track friction
conditions. Using basic mechanics and resolving forces leads to the following equation:
X
Forces = ma (9.7.1)
where F is the force, m is the mass, g is gravitational acceleration, θ is the angle of the incline and
a is the acceleration. Inputting the appropriate values into Equation 9.7.2 leads to:
X
Ftrailing max = 600 × 10 sin 2.86 + 600 × 2.0 = 1499.37...N ≈ 1500 N. (9.7.3)
Based on Equation 9.7.3, a rearward acting force of 1500 N on the underside of the rear double-
height frame beam was added to the previous simulation (including generator and component
loads). Figure 127b shows a von Mises stress plot for the trailing load simulation, with a maximum
stress of 0.162 GPa representing a factor of safety of 1.33 over the yield stress value. It should be
reiterated that the conditions for the maximum acceleration as predicted in this study are incredibly
9.8. FINAL DESIGN 167
unlikely as track friction has to be incredibly high. Regardless, the worst case simulation demon-
strates the suitability of the frame for purpose. Figure 127c shows the displacement plot for the
trailing load scenario with maximum displacement of around 1.87 mm encountered at the drawbar
connection point at the application of the trailing load, as is to be expected.
(a) FEA setup - note the addition of the 1500N force acting centrally on the underside of the foremost double
height beam
Figure 127: Frame FEA study for locomotive at rest with 1500 N backward-acting force to simulate
maximum acceleration up a 2.86° slope with a 600 kg trailing load
10. Body Shell
10.1 Requirements CM HMF
Factors such as shape, access, ventilation, material, and safety should be taken into consideration
when designing the shell. Shell dimensions are heavily influenced by the shape selected, along
with frame size. Details such as points of access to internals, ventilation, and waterproofing can
help prevent failure and increase functionality. Implementation of noise control methods can also
influence competition performance.
Mitigating steps can now be carried out to reduce the effectiveness of each noise transmission
path; airborne and structural based vibration. If either of the two transmission paths were to be in-
terrupted then the noise magnitude arriving at the microphone would be reduced. All steps taken to
minimise the noise escaping the shell are passive, and depend on structural design considerations.
Two different approaches can be used to control airborne based transmission paths - reflection
and dissipation [148]. Reflection occurs when a wall or lining obscures the passage of sound,
and returns it back from its surface. Dissipation occurs when the surface produces heat from the
absorbed sound energy, although in quantities negligible compared to the heat already emitted
from sources within the locomotive systems. Ideally a combination of the two sound reduction
169
170 CHAPTER 10. BODY SHELL
mechanisms are implemented to maximise the sound level reduction, however, modifications made
to reduce sound emissions should not impede locomotive functionality.
Different porous materials such as sound absorbent foam can be applied to interior surfaces
of the locomotive to provide sound reduction ability through dissipation. Of course these materials
have to be appropriate for the intended use (materials should be fire and heat proof), and this limits
the choice available. This method can be subsequently investigated for use in future years.
In principle any shell around the body of the locomotive will induce both reflection and dissipa-
tion of sound energy and hence, the sound levels experienced at the microphone.
To achieve adequate ventilation, incorporation of air vents, or cut-outs, in the shell are required.
Breaks in material continuity such as air vents, or open access panels, can compromise the noise
reducing capabilities of the shell, although intelligent design decisions can be made to reduce this
effect. A simple example of such design decisions include the use of louvred air vents (Figure 130b)
over slotted air vents (Figure 130a). This is because the louvred air vents reflect sound towards the
ground, due to their opening orientation, whilst maintaining adequate ventilation ability.
(a) Slotted air vent [149] (b) Louvred air vent [150]
Noise transmission pathways occur with structural objects carrying the vibrational energy. This can
then lead to vibration of other objects within the construction, resulting in noise being produced
from the vibrational movement of the part. This can be mitigated and avoided as much as possible
by careful design considerations to avoid excessive part vibration, and damping should it occur.
Structural damping is achieved within the shell by incorporation of rubber intermediate lining
between plates, which act to convert vibrational energy to thermal, removing it from the system.
−2
10
−3
10
FFT Magnitude, V
−4
10
−5
10
1 2 3
10 10 10
Frequency, Hz
(a)
Setup of the microphone to calcu- (b) Magnitude of receiver voltage at different frequen-
late the noise levels cies
Fourier transform operations carried out on the microphone output produce a spectra of relative
amplitudes at given frequencies, this can be seen in Figure 131b. Inspection of Figure 131b shows
resonance at various frequencies, which can be identified by sharp peaks. This recording is taken
from the generator running at full power in front of the exhaust.
Figure131b shows resonance at 35.5 Hz, 71.1 Hz and 106.5 Hz - all multiples of 35.5 Hz.
The resonant frequencies listed are dependent upon generator cycle frequency; where the lowest
natural frequency (35.55 Hz) can be associated with the ignition frequency of the generator, and
the second lowest frequency the rotational frequency of the engine (71.1 Hz), a running speed of
4266 RPM.
Further sound recording analysis yields the generator SWL spectra. A microphone calibration
factor is applied to the sum of the SPLs obtained from five different locations. This gives the SWL
of the generator which can be seen in Figure 132, which shows the generator A-weighted SWL
(dB), non A-weighted SWL (dBA) and background noise plots (dB). The A-weighting is applied to
sound spectra to simulate loudness as perceived by the human ear at specific frequencies [151].
Figure 132: A-weighted and non A-weighted SWLoof the generator compared to the SWL of back-
ground noise [152]
172 CHAPTER 10. BODY SHELL
It can be seen that relative to the total noise produced by the generator, the background noise is
insignificant. In the A-weighted SWL, the most significant noise contribution is due to the frequen-
cies between 400 - 4000 Hz. The intention is to reduce noise output of the locomotive, concentrat-
ing on this frequency band, with the addition of a shell.
The sound reduction index has been calculated for shells of different materials, and thickness;
1.5 mm aluminum, 3 mm aluminum and 3 mm fiberglass. Relative performances of each shell can
be seen in Figure 133. The SPLs at the microphone are calculated assuming that the generator
acts as a point source, and that the noise spreads as a hemisphere over rigid ground. The SPL at
the microphone for each shell is shown in Figure 134, along with a reference SPL in which no shell
is present. This was achieved utilising code adapted from a previous assignment: Noise Control
Engineering (ISVR 3064) [152].
Overall model results are displayed in Table 21, detailing modelled SPLs experienced by the
microphone, and the resonant frequencies of the shell panels.
70
3.0 mm Al
1.5 mm Al
60 3mm Fiber Glass
50
Reduction Index,(dB)
40
30
20
10
0
2 3 4
10 10 10
Frequency (Hz)
Figure 133: Reduction index for the three shell options [152]
80
1.5 mm Al
3 mm Al
70
Sound Pressure Level Reciever (dBA)
No Shell (Control)
Fiber Glass 3mm
60
50
40
30
20
10
1 2 3 4
10 10 10 10
Frequency (Hz)
Figure 134: One third octave band spectra SPL at receiver with different shells [152]
The model predicts a minimum SPL drop of 20 dBA, with particular shells performing better
than others. 1.5 mm aluminium yields the highest SPL reduction compared to the control.
Changing the thickness of aluminium from 1.5 mm to 3 mm increases the overall noise level
because the critical plate frequency drops from 5466 Hz to 2733 Hz. This has a negative impact
because 2733 Hz is within the region of the high amplitude frequencies present, whilst 5655 Hz
10.3. MATERIAL CHOICE 173
Table 21: Critical frequencies and SPLs at the receiver for various shells
is above the this dominant frequency band. Fibreglass can be seen to be the least favourable
acoustic insulation material.
The noise reductions predicted are likely to be excessive, and in reality will be of a lower magni-
tude. This is due to simplifications of a complex situation, including the omission of certain features
such as ventilation access, and the open locomotive base. Due to the omitted features being con-
stant across each shell choice, it is likely that the relative performance of each shell holds true with
the predicted results. Therefore, the model can be used as a useful indicative tool for finding the
favourable shell design. In this case, it indicates that the most appropriate choice of shell material
for sound insulation purposes is 1.5 mm thick aluminium.
without the need for additional training. Composite manufacture requires external aid. Fortunately,
a Southampton University graduate owns and operates his own composite fabrication facility, and
has offered his services. This option offers the additional opportunity for the group to gain hands-on
composite manufacturing skills.
For sheet metal, aluminium was determined to be the best choice, primarily due to its high
specific modulus [154], ease of manufacturability, and favourable sound insulation properties. Most
grades of aluminium would be sufficient, , but due to being work hardened, the 5xxx series exhibits
Lüders elongation bands (stretch marks) which cannot be painted over and so this grade should
be avoided. [155].
Fibreglass is an appropriate choice if the shell is to be moulded. It is a relatively common
composite material with a proven history in structural shell-type environments such as boat hulls
[156]. It has a low density, and by using wet lay-up manufacturing techniques, its thickness is
controllable to a low tolerance [157].
Cost
In talks with the Southampton University alumnus, a quote was negotiated to construct one com-
plete shell (painted, vented, and with access panels) for £1500.
Based on frame dimensions of 2000mm × 600mm, and shelf height of 600 mm, a maximum of
three sheets of aluminium would be required to cover the locomotive adequately. At a rough cost
of between £30 - £50 per sheet (depending on thickness), three sheets should cost no more than
£150. As machining the sheets will be done ourselves, manufacturing costs would be negligible.
The other major costs attributed to a sheet metal shell are: the nuts and bolts required for shell-
frame interaction; sealing material (such as rubber) to provide waterproofing; and painting. These
expenses are approximately equivalent to the cost of the sheet metal. As a result, the sheet metal
shell is significantly cheaper to the fibreglass alternative.
Due to the lower cost and other consideration in Section 10.3.1, aluminium is the obvious shell ma-
terial choice. This decision is consistent with previous entries into the IMechE Railway Challenge,
of which the vast majority have also opted for aluminium constructions. Another deciding factor is
10.4. DESIGN DECISIONS 175
that aluminium is recyclable, whereas fibreglass is not. Competition entry requirements state that
95% of the locomotive components (by mass) must be recyclable. With a target locomotive mass
of 500 kg, only 25 kg of the train can be non-recyclable. Using a fibreglass shell would take up a
large proportion, if not all, of that allowance.
Most of the aluminium shell is 1.5 mm thick. Upon physical inspection of the material, it did
not collapse or deform under its own weight, and has been theoretically shown to provide superior
sound insulation than thicker sheets (see Section 10.2).
10.4.1 Weatherproofing
As with any vehicle, the locomotive must be proofed against inclement weather conditions that
could damage system functionality. The locomotive will not be expected to function in extreme con-
ditions but will be expected to operate in either heavy rain or high temperatures. Weatherproofing
measures taken in the shell design to ensure it is able to do so adequately are outlined below.
Waterproofing
To protect equipment that would be damaged should water gain access to the locomotive interior, a
number of design features including overhangs, rubber seals, louvred vents and tight dimensional
constraints are implemented.
Wherever a transition from the roof sheet to the side sheets of the shell exist, an overhang has
been used to eliminate the chance of water entering the locomotive between gaps in the panels.
This can be seen in Figures 135a and 135b, which show the two different overhang types used.
The design encourages the flow of water to be diverted outside of the aluminium side panels and
onto the floor. A 10 mm opening exists where the front and rear shell sections meet the roof of
the shell for ventilation purposes. Water access is restricted here by a larger 30 mm overhang
perpendicular to the opening, as seen in Figure 135a. This is in contrast to the overhang running
either side of roof, along the longitudinal length of the locomotive, which is tightly constrained to
allow neither ventilation or water access, as can be seen in Figure 135b. These plates are bolted
onto the frame, with a 45 mm overlap between the two sheets.
176 CHAPTER 10. BODY SHELL
(a) Front overhang 30 mm long, covering 10 mm wide (b) Side overhang of roof (blue) tightly
opening constrained to side panel (red)
The louvred venting system used on the locomotive side panels also utilises overhanging edges
to retard water access, encouraging water run-off instead (Figure 130b). Tight waterproof seals are
also implemented where two sections of sheet metal meet. The rubber seals act as a hydrophobic
gap sealant.
For operation at high ambient temperatures, the locomotive shell has been designed in a modular
manner so that the system is capable of running without the roof, or part of the side panel assembly.
In addition to being able to provide additional cooling to contained components, running with certain
panels off also allows ease of access and insight to the internal running of the locomotive. To
ensure the necessary safety measures are still in place when this convenience feature is being
utilised, the emergency stop buttons are placed on the panels of the shell which are least likely to
be removed during running.
Enabling Access
In order to accommodate the access requirements detailed above, the entire shell is removable
in individual elements due to its modular design, though some components need others to be
removed before they can be. This allows general access to all areas on the locomotive, and due
to the manageable size of each part, this can be achieved by an individual. Furthermore, the side
panels are assembled as three individual modules on each side, and are assembled in a way to
10.4. DESIGN DECISIONS 177
allow easy access to the generator from one side and the shelves from the other side (Figures 136
and 137). To allow the outermost plates to be removed without the others coming loose, the outer
plates incorporate oversized holes through which inner plates are independently secured.
Several features of the shell require totally unrestricted access at all times, including the emergency
stop button and the accelerometer mounting point. One emergency stop button is mounted on
on the innermost plate each side of the locomotive. The IMechE specifications require that a
25 mm × 25 mm ferrous plate is mounted with unrestricted access on the locomotive structure [15].
The location chosen for this is on top of the roof, attached using a bolt through the aluminium sheet
into the extruded beam beneath.
Figure 136: Right side of locomotive displaying access to generator housing with middle panel
removed
Figure 137: Left side of locomotive displaying access to shelving units with end panels removed
The ventilation openings in the front and back panel provide a through-flow of air when the lo-
comotive is in motion, extracting heat energy generated within the locomotive. Due to the tendency
of heat to rise, openings at the top of the locomotive allow the hot air to exit.
To encourage heat exchange, the bottom of the locomotive is left open. This enables the domi-
nant air flow regime to develop when in motion, be it bottom to top, along the top from front to back,
or a mixture of the two. When stationary the top openings will not act as an air intake, rather they
will act as an exhaust for the air heated within the shell which then goes on to rise via convection,
forming a passive cooling cycle of air entering the bottom, and leaving via the top openings.
Additional ventilation points exist on the side panels in the form of louvred vents (see Figure
130b), supplying location specific openings to allow fresh air and heat exchange. The locations
coincide with the shelving units and the generator housing, with one vent on each side panel,
totalling six.
(a) Siemens 444 train at Northam depot (b) Siemens 450 train at Eastleigh train station
The overall shape for the two trains seems almost identical. The main difference seems to
be the actual positioning of the doors on the train - one optimised for long-haul express journeys
(Siemens 444), the other for stopping services (Siemens 450), and the colour scheme. As the
locomotive in this project has no good reason to have train doors similar to that of the full-size
Siemens train, this minor subtlety is of little concern. The front of the two models, however, look
identical, which will be the only paint scheme replicated as close to accurately as possible. The
sides have sponsorship logos along with features such as venting and safety buttons.
(a) Lab space utilised to construct the loco- (b) Actual 3 m long, 10.25" gauge track pan-
motive els used for the project
Lockable storage for both parts and tools were also arranged by the lab supervisor Tim Hartley.
Due to the anticipated competition participation, the team were also required to obtain all the nec-
essary tools for both building and fixing of the locomotive where required. This does, however, set
up the project incredibly well for future years should it be rerun, which it almost certainly will be.
180
11.2. BUILD RIG 181
With the positions of loading roughly known, the next consideration for build rig construction
was connector strength for the aluminium frame. The gussets (right angled connectors) selected
182 CHAPTER 11. BUILD AREA AND RIG
for the build frame were the standard 45 mm Bosch Rexroth gussets. The maximum loading of the
gussets as recommended by Bosch Rexroth are indicated in Figure 120a on page 161. Assuming
a worst case loading of 6 kN spread across the four sleepers and so through eight upright gussets,
it was clear even a small moment arm would load the gussets dangerously close to the maximum
recommended bending moment. Based on the recommended loadings of the gussets, efforts were
made to load the beams beneath the sleepers in such a way that the gusset load was minimised.
The build rig concept that was designed is shown below in Figure 142 in various loaded and
unloaded conditions.
In order to validate the build rig as indicated in Figure 142, FEA was carried out. The geometry
was broken down into elements to form a mesh, loading conditions, connections and fixtures were
then set and the results analysed. Rigid connections were used to simulate the gussets as this
is the closest match to their mechanical behaviour. Split lines were drawn onto the faces of the
beams which would contact the gussets in order to more closely replicate the gusset behaviour.
Initially FEA was attempted on the entire build rig geometry as indicated in Figure 142f. The mate-
rial specified for all beams was Aluminium 6063 T6 as indicated by the frame catalogue technical
specifications [147]. The purple arrows represent the load condition which was set to a worst case
total of 6 kN combined across the loaded sleepers, the green arrows represent the fixtures which
in this case would be the floor mounts on the lower face of the rig. Blue arrows represent the rigid
connections modelled between the frame faces to replicate the behaviour of the gussets. Unfortu-
nately problems were encountered when meshing such a large assembly. The main issue was that
the manufacturer-provided Bosch Rexroth aluminium profile CAD features the intricate geometry of
the inner sections, requiring a fine mesh in order to successfully mesh certain beams. Combined
with such a large geometry, the required fine mesh proved a problem for the limited computing
power both using university computers and personal computers at home. Due to regularly exceed-
ing the available memory of the machines and general SolidWorks instability, a compromise was
sought.
In order to still be able to simulate the build rig under the loading conditions, the rig geometry
was split into quarters. As the loading is symmetrical across the rig, this seemed a fair compromise.
It should be noted that there will be differences in the exact stress/displacement figures as a beam
which is simply supported at both ends deflects less than a cantilever beam of half the length.
As such, the build rig geometry was split into quarters. Figure 143a indicates the FEA setup with
the loading having been reduced to 1.5 kN total across the purple arrows. Figure 143b and 143c
indicate the results of the study. Figure 143b, showing a von Mises stress plot of the assembly,
indicates a factor of safety of around 2.43. It should be remembered that the simulated results will
be distinctly worse than the real life performance of the entire frame and, as such, the build frame
was validated as fit for purpose. In order to confirm that splitting the structure into quarters would
still provide safe results, the structure was modelled using beam theory. For a cantilever beam of
length L subjected to a load W , Young’s Modulus E and second moment of area I the maximum
deflection is given by [37]:
WL3
νmaximum = (11.2.1)
3EI
and maximum bending moment, BM is given by [37]:
(a) Isometric view of build rig geometry without (b) Right view of build rig geometry without
sleeper and locomotive sleeper and locomotive
(c) Isometric view of build rig geometry with (d) Isometric view of build rig geometry with
sleeper and locomotive sleeper and locomotive
For a beam which is fixed at both ends and subject to a central loading of W, the maximum deflec-
tion is given by [37]:
WL3
νmaximum = (11.2.3)
192EI
and maximum bending moment, BM is given by [37]:
WL
BMmaximum = . (11.2.4)
8
In order to simplify the FEA, it was proposed that the overall structure be split into quarters such
L W
that the width-ways support beams were of length 2 and the load was reduced to 2 . In the
simplified structure the beams were modelled as cantilevers with half the length and half of the
load. Substituting these values into 11.2.1 and 11.2.2 leads to:
W L 3
2 (2) WL
νmaximum = = (11.2.5)
3EI 48
W L WL
BMmaximum = = . (11.2.6)
2 2 4
Relating stress σ, bending moment BM, thickness y and second moment of area I by using [37]:
BMy
σ= (11.2.7)
I
it is observed that for constant values of y and I, the stress σ is linearly proportional to the bending
moment BM. Comparing the cantilever deflection and max bending moment for the simplified case
as in Equations 11.2.5 and 11.2.6 with the actual cases as in Equations 11.2.3 and 11.2.4 it is
easy to see that the actual case has a greatly reduced maximum deflection (factor of four) and
maximum bending moment (factor of two), and so stress. As such, it can be assumed that if the
simplified FEA structure is proven safe, the actual structure is more than sufficient for the required
loadings.
(a) 3D printed aluminium profile cover (b) End profile cover in place on build rig
Figure 143: Build rig FEA study having split the geometry into a quarter of actual layout
12. Design Summary MC
Based on the combined efforts of all team members, the project has successfully delivered a func-
tioning miniature locomotive (Figure 146) with the means to recover energy from the braking pro-
cess. A short summary of each subsystem is detailed below (further details can be found in the
respective technical chapters):
• Bogies Two identical bogies, each featuring two driven wheelsets (Figure 147), have been
designed and assembled. The wheels are 250 mm in diameter, manufactured from EN8
(medium carbon steel) and dimensions as specified by the IMechE. Each wheelset features
a multi-spline axle to aid design flexibility in future years, two wheels, a brake disc and a 50
tooth sprocket for driving. Integration of brake and traction systems has been carried out
along with the attachment of the bogies to the main locomotive frame by a flanged turntable
bearing. A primary suspension system comprising two coil springs per bearing adapter pair
(or four coil springs per wheelset) has been implemented to reduce vibration. See Figure 148.
• Brakes A vacuum brake system has been specified, designed, tested and integrated into
the bogie design along with a means to manually release and reapply the brakes in accor-
dance with the IMechE Railway Challenge requirements by using a hand-held vacuum pump.
Solenoids in the brake loop, controlled by the myRIO, are used to actuate the brake cylinder
which is connected to mechanical go-kart brakes by a mechanical linkage.
• Electronics & Control A wide range of electronics and control systems across a number
of other subsystems have been modelled, designed, tested and integrated throughout the
entire locomotive. A large number of parts have been selected, purchased and implemented
including the selection of a suitable motor controller with an energy recovery mode of op-
eration. A National Instruments myRIO microcontroller has been programmed in LabVIEW
to connect the various subsystems together. Locomotive control occurs remotely by a Wi-Fi
link between a tablet running a National Instruments Dashboard app and the Wi-Fi hotspot
myRIO. Subsystems controlled include:
– Sensing (fire detection in the form of heat sensors, speed monitoring by use of a hall
effect sensor)
– Brake system including control of the vacuum pump, solenoid actuation, pressure sens-
ing
– Energy recovery including the charging of the supercapacitor by switching the mode of
operation of the motor controller, monitoring of the charge level of said supercapacitor
and subsequently discharging the supercapacitor across the motor terminals
– Emergency stop systems
– Traction including motor controller speed variation, swapping of drive/energy recovery
modes and limiting speed in accordance with IMechE requirements.
• Energy Recovery A supercapacitor system allowing for the energy recovered from the wheels
during braking to be stored for use at a later time to drive the locomotive forwards has been
implemented.
• Frame A ladder frame has been designed and calculations & FEA studies carried out to
confirm suitability for the mounting of all other components including bogies and shell. A
prototyping aluminium grooved frame has been selected in order to make future modifications
187
188 CHAPTER 12. DESIGN SUMMARY
to the frame simpler. Shelving units have been added to the frame for storage of electronic
components such as batteries, motor controller and supercapacitor.
• Shell An aluminium plate shell exterior has been designed to bolt onto the existing frame
shelving units. It is envisaged the plates will be powder coated for aesthetics and sealed for
weatherproofing.
• Traction Two Lynch LEM170 DC motors were selected for use based on numerical models
of power requirements of the locomotive and a review of alternative options. A monomotor
design sharing one motor between two wheelsets has been designed and geared at a 5:1
ratio to reduce the motor RPM to a more suitable locomotive speed in-line with the technical
requirements. A petrol generator was selected for use as the primary energy source and
efforts made to isolate the vibrations throughout the locomotive.
Figure 147: Close-up of locomotive wheelset showing driven sprocket and disc brake
189
190
13.2. LOCOMOTIVE TEST RUN 191
(a) The locating sleeve used to constrain the (b) Drawbar as fabricated for coupling at
drawbar Eastleigh
Soft Charge PN
Part of the testing included using the soft charging circuit for the supercapacitor. The SMPS output
a voltage of 40.6 V through a 1 Ω resistor. Once the supercapacitor voltage reached 37 V the
soft charge resistor was bypassed. The recorded results of this test (Figure 153a) were compared
with those from a representative Simulink model (Figure 153b) and found to be almost identical.
Simulations provide safe testing, and future work can now be undertaken to identify the optimal
state of charge at which the resistor can be bypassed.
During testing the electronics & control systems on board the locomotive were all largely functional.
A number of observations were made during testing including:
• The integrated system, which had been bench tested before integration into the locomotive,
performed as intended
• It was observed that the supercapacitor soft charge resistor could be bypassed in order to
reduce the time taken to charge said supercapacitor
• The myRIO could not be programmed such as to run the VI independently of the laptop and
so the laptop had to remain connected for the duration of testing
• No fuses were blown, circuits shorted or cables cut, suggesting the overall system is safe in
operation as expected
• The two lead-acid batteries were used over both days with sufficient charge, confirming they
are suitable for the system requirements.
Mechanical Observations
As well as the electronics & control aspects of the locomotive, testing provided the opportunity to
make a number of observations around the mechanical side of the project. These included the
following points:
• It was noticed that the current suspension arrangement (springs located too close to one
another and the lockout plate slots insufficiently constraining the wheelsets) lead to twisting
of the wheelsets relative to the frame mounts. This is recommended as an area for future
work.
• During testing problems were encountered with the stretching of the chain between motor
and axle sprockets. As the sprockets used in testing were not the sprockets the locomotive
was designed for, the team are confident such issues will be alleviated once the intended
sprockets are installed.
• Mechanical systems generally performed well, as anticipated
• One of the brake calliper mounts visibly deformed after use with a larger-than-expected trail-
ing load. This is likely due to a reduced number of operating brake callipers during testing,
compared to the number intended for use, leading to larger braking forces transmitted through
the bracket.
• The motor mount was observed to twist relative to its mounting position under tension from
the drivetrain, indicating further constraint in the attachment is required
• The bogie-frame bearings, which had a concerning amount of play during assembly, were
sufficiently constrained by using nylon blocks at either side of the frame where said bearings
were mounted.
Overall, locomotive testing at Eastleigh Railway was a real success with systems, generally, per-
forming as intended. The testing provided the opportunity for the team to observe the performance
of the locomotive on a purpose-built railway line and areas for further work, in preparation for the
Railway Challenge, were identified as a result. Figure 154 shows some images of the locomotive
during testing.
194 CHAPTER 13. TESTING
• Bogies
– Shell manufacture
– Testing and optimisation of noise damping performance
• Traction
– Fit the ordered (but not received at time of submission) sprockets and so final drivetrain
system
– Optimise acceleration rates using the motor controller.
195
196 CHAPTER 14. FUTURE WORK
• Bogies
(a) CAD concept model of Siemens 450 (b) CAD concept model of Siemens 444
class train class train
• Traction
– Investigate plausibility of using a fuel cell as the primary energy source of the locomotive
for greatly improved noise and vibration performance.
15. Conclusions MC
A 10¼" gauge miniature locomotive incorporating a number of subsystems has been successfully
designed, manufactured, assembled and tested to comply with a technical specification as provided
by the IMechE for the Railway Challenge competition.
There have been number of issues which the team have encountered throughout the project,
namely:
• Lead times for components which were much longer than expected
• Lead times and import charges for components ordered from overseas
• Considerable lead times for work which was submitted through the EDMC (parts submitted
mid-December were finished mid-March)
• Delays and extra expense incurred as a result of the axles complicated keyway design which,
in hindsight, should have been a simple ISO keyway but wasn’t known about at the time
• Minor repetition of work where team members’ activities are not communicated sufficiently
clearly to the rest of the team
• Manufacturing issues due to inexperience of team members in having designs manufactured.
Tolerancing and machining capability in particular.
• Considerable delays as a result of ordering components through Siemens (industrial spon-
sor). Notably an RS order which took around 10 weeks to arrive and a sprocket order which
took around 6 weeks to arrive, both of which should have arrived within a maximum of 2
weeks.
As one of the largest GDP groups, the team have successfully worked together in a limited time
frame to construct a functional locomotive. Further work has been recommended for activities after
the report deadline in preparation for the Railway Challenge competition, and also for next year’s
project team. The efforts of the current team have resulted in a functional and flexible initial locomo-
tive, allowing subsequent teams to optimise the existing design and control systems. Infrastructure
such as the means to safely raise the locomotive to an ergonomic height for construction, ability
to move the locomotive once at floor level and track panels to allow local testing of the locomotive
have all been designed or acquired as needed. Strong connections have been forged with: East-
leigh Lakeside Railway for testing & advice; the IMechE for subsequent competition entries and
local companies such as Clarke Lane Engineering. As well as attracting attention across the whole
of the engineering faculty, the project has drawn particular interest from the Dean of Engineering,
Professor William Powrie, been used for a science and engineering open day in collusion with the
Track 21 project and will be shown at the inaugural faculty design show in June. Finally, the team
successfully applied for and won a “rookie team grant” from the IMechE (for entry into the Railway
Challenge) to the value of £1,500 which is to be used in addition to the basic GDP budget for next
year’s team.
The project has provided a fantastic learning experience for all involved and the team both hope
and expect that the optimisation of the current miniature locomotive design will exist as a GDP for
many years to come.
197
Bibliography
[1] C. Wolmar, Fire and Steam: A New History of the Railways in Britain. Atlantic Books,
Limited, 2009. [Online]. Available: https://books.google.co.uk/books?id=4xkAIW_4ydAC
[2] J. Harvey, N. Thorpe, M. Caygill, and A. Namdeo, “Public attitudes to and perceptions of
high speed rail in the UK,” Transport Policy, vol. 36, no. 0, pp. 70 – 78, 2014. [Online].
Available: http://www.sciencedirect.com/science/article/pii/S0967070X14001541
[4] D. MacKay, Sustainable Energy without the Hot Air, ser. Without the Hot Air Series. UIT,
2009. [Online]. Available: https://books.google.co.uk/books?id=IsgiPQAACAAJ
[5] Ten and a Quarter Inch Railway Society. UK miniature railways. [Online]. Available:
http://www.tenquarter.org/public-railways/
[6] Eastleigh Lakeside Steam Railway. History and line. [Online]. Available: http://www.
steamtrain.co.uk/history
[9] J. A. Lozano, J. Félez, J. d. Dios Sanz, and J. M. Mera, Railway Traction. InTech.
[12] Apogee Interactive Inc. (2013) Electric Motor Advantages and Disadvantages. [Online].
Available: http://c03.apogee.net/contentplayer/?coursetype=md&utilityid=elpaso&id=12539
[13] T. F. S Iwnicki. (2015, Mar) Institution of mechanical engineers railway challenge rules.
[14] R. Bansal, A Textbook of Fluid Mechanics. Laxmi Publications Pvt Limited, 2005. [Online].
Available: https://books.google.co.uk/books?id=FzQz6A6SnyoC
[15] T. Poole. (2015, feb) Institution of mechanical engineers railway challenge technical spec-
ification. [Online]. Available: http://www.imeche.org/docs/default-source/railway-challenge/
railway-challenge-2015-technical-specification-issue-1.pdf
[16] C. Barkan. (2007) Railroad transportation energy efficiency. [Online]. Available: http:
//www.istc.illinois.edu/about/SeminarPresentations/20091118.pdf
198
BIBLIOGRAPHY 199
[17] Railway Technical Pages. (2014) Electric traction drives. [Online]. Available: http:
//www.railway-technical.com/drives.shtml
[19] Robot Marketplace. (2014) Image of Etek Motor. [Online]. Available: http://www.
robotmarketplace.com/products/images/store_brushlessetek_lg.jpg
[20] Robot Marketplace. (2014) Image of Perm Motor. [Online]. Available: http://www.
robotmarketplace.com/products/images/store_permmotors_lg.jpg
[22] Renold. Wheels and Sprockets: Designer Guide. [Online]. Available: http://www.renold.com/
UploadedFiles/Brochure_Sprockets.pdf
[23] Utility Free Living. (2015) Image of UFLDC3000S. [Online]. Available: http://www.
utilityfreeliving.co.uk/wp-content/uploads/2015/01/Silent-3kW-DC-Generator-named-3.png
[27] H. Poincaré, Science and Hypothesis. Walter Scott Publishing Co. Ltd, 1952.
[29] Flybrid Automotive Limited. (2014) Original f1 system. [Online]. Available: http:
//www.flybridsystems.com/F1System.html
[32] National Academy of Sciences. (2008) What you need to know about energy. [Online].
Available: http://www.nap.edu/reports/energy/sources.html
[33] Y. Gogotsi and P. Simon, “True performance metrics in electrochemical energy storage,”
Science Magazine, vol. 334, no. 6058, pp. 917 – 918, 2011. [Online]. Available:
http://www.sciencemag.org/content/334/6058/917
[34] D. Ragone, “Review of battery systems for electrically powered vehicles,” SAE, Tech. Rep.
200 BIBLIOGRAPHY
[35] H. L. H. Satish Samineni, Brian K. Johnson and J. D. Law, “Modeling and analysis of
a flywheel energy storage system for voltage sag correction,” IEEE TRANSACTIONS
ON INDUSTRY APPLICATIONS, vol. 42, no. 1, pp. 42 – 52, 2006. [Online]. Available:
http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=1583828
[37] J. R. Calvert and R. A. Farrar, An engineering data book, 3rd ed. Basingstoke: Palgrave
Macmillan, 2008.
[38] J. Hampl, “Concept of the mechanically powered gyrobus,” Transactions on Transport Sci-
ences, vol. 6, no. 1, pp. 27 – 38, 2013. [Online]. Available: http://www.degruyter.com/view/j/
trans.2013.6.issue-1/v10158-012-0029-9/v10158-012-0029-9.xml?rskey=beFtyU&result=2
[40] S. Nathan. (2012) Grid-connected energy storage: a new piece in the uk en-
ergy puzzle. [Online]. Available: http://www.theengineer.co.uk/in-depth/the-big-story/
grid-connected-energy-storage-a-new-piece-in-the-uk-energy-puzzle/1014536.article
[41] A. M. Saurabh Singal, Kapil Parikh, “Design and simulation of flywheel energy
storage an alternative to ups batteries,” Progress in Science and Engineering,
vol. 2, no. 6, pp. 52 – 61, 2014. [Online]. Available: http://piserjournal.org/2014/11/
design-simulation-of-flywheel-energy-storage-an-alternative-to-ups-batteries/
[42] Synchrony. (2015) How magnetic bearings work. [Online]. Available: http://www.synchrony.
com/knowledge/how-magnetic-bearings-work.php
[48] J. Zhang, L. Zhang, H. Liu, A. Sun, and R. Liu, Electrochemical Technologies for Energy
Storage and Conversion. Wiley, 2012. [Online]. Available: http://books.google.co.uk/books?
id=AN3B3L5RtqUC
BIBLIOGRAPHY 201
[50] S. Liu, L. Xiong, and C. He, “Long cycle life lithium ion battery with lithium nickel cobalt
manganese oxide (ncm) cathode,” Journal of Power Sources, vol. 261, no. 0, pp. 285 – 291,
2014.
[53] R. K “Principles and applications of electrochemical capacitors,” Electrochimica Acta, vol. 45,
no. 15âĂŞ 16, pp. 2483 – 2498, 2000.
[54] M. T. Bobby Maher. Ultracapacitors provide cost and energy savings for public transportation
applications. [Online]. Available: http://www.batterypoweronline.com/images/PDFs_articles_
whitepaper_appros/Maxwell%20Technologies.pdf
[55] “Integrating large specific surface area and high conductivity in hydrogenated NiCo2 O4
double-shell hollow spheres to improve supercapacitors,” NPG Asia Materials, vol. 7, no.
165, 2015.
[57] M. Wens and M. Steyaert, Design and Implementation of Fully-Integrated Inductive DC-DC
Converters in Standard CMOS, ser. Analog Circuits and Signal Processing. Springer,
2011. [Online]. Available: http://books.google.co.uk/books?id=KwKv_99mmYMC
[58] C. Hubert, Electric circuits AC/DC: an integrated approach, ser. McGraw-Hill series in
electrical engineering: Electronics and electronic circuits. McGraw-Hill, 1982. [Online].
Available: http://books.google.co.uk/books?id=L-FSAAAAMAAJ
[59] T. Longland, T. Hunt, and W. Brecknell, Power capacitor handbook. Butterworths, 1984.
[Online]. Available: http://books.google.co.uk/books?id=u-RSAAAAMAAJ
[61] A. Ltd. (2015) Alumina-aluminium oxide-al2o3-a refractory ceramic oxide. [Online]. Available:
http://www.azom.com/properties.aspx?ArticleID=52
[63] MakerBot. PLA and ABS strength data. [Online]. Available: https://eu.makerbot.com/
fileadmin/Inhalte/Support/Datenblatt/MakerBot_R__PLA_and_ABS_Strength_Data.pdf
[66] 4QD. (2015) Instruction manual: 4qd series controllers. [Online]. Available: http:
//www.4qd.co.uk/instrs/New/View/4QD_300.pdf
[67] Department for Transport. Intercity express program schedule 1 - appendix a - technical
specification. [Online]. Available: https://www.gov.uk/government/uploads/system/uploads/
attachment_data/file/82840/tts-redacted.pdf
[68] J. H. White, Jr., The American Railroad Passenger Car: Part 2), ser. Johns Hopkins Studies
in the History of Technology. The Johns Hopkins University Press, 1985.
[69] S. P. Sutera and R. Skalak, “The history of poiseuille’s law,” Annual Review of Fluid Mechan-
ics, vol. 25, no. 1, pp. 1–20, 1993.
[70] K. Fender. South west trains upgrades emu fleet. [Online]. Available: http://www.railjournal.
com/index.php/rolling-stock/south-west-trains-upgrades-emu-fleet.html
[72] Railway Technical Web Pages. (2014) A guide to airbrakes. [Online]. Available:
http://www.railway-technical.com/air-brakes.shtml,
[73] Take-y. (2006) Eddy current brake equipment on shinkansen series 700. [Online]. Available:
http://commons.wikimedia.org/wiki/File:Uzu-brake.JPG
[74] M. Fujita, T. Tokumasu, T. Yamada, T. Hirose, Y. Tanaka, N. Kumagai, and S. Uchida, “3-
dimensional electromagnetic analysis and design of an eddy-current rail brake system,” Mag-
netics, IEEE Transactions on, vol. 34, no. 5, pp. 3548–3551, Sep 1998.
[77] P. Wang and S. Chiueh, “Analysis of eddy-current brakes for high speed railway,” Magnetics,
IEEE Transactions on, vol. 34, no. 4, pp. 1237–1239, Jul 1998.
[78] E. Draper. (2008, Aug) Air brakes on the Golden Valley Light Railway. [Online]. Available:
http://www.gvlr.org.uk/tech_info/airbrakes.pdf
[79] Railway Technical Web Pages. (2014) A guide to vacuum brakes. [Online]. Available:
http://www.railway-technical.com/vacuum.shtmll,
[80] Bridge End Minature Railway. (2014) Locomotives. [Online]. Available: http://www.
bridgendminiaturerailway.com/locomotives.html
[82] L. E. Z. J. Reuter, D. and J. Elliott, “Hydraulic design considerations for EHB systems,” SAE
Technical Paper.
[85] The Ergonomics Centre. (2006) Human strength data tables. [Online]. Available:
http://www.theergonomicscenter.com/graphics/Workstation%20Design/Strength.pdf
[86] M. Tomorrow. (2013, June) History of the microprocessor. [Online]. Available: http:
//meetingtomorrow.com/cms-category/history-of-the-microprocessor
[87] S. Sone and I. Okumura. (1983) The microprocessor in railway control systems. [Online].
Available: http://link.springer.com/chapter/10.1007%2F978-94-009-7007-6_19
[88] Arduino. (2015) Arduino boards and products. [Online]. Available: http://arduino.cc/en/Main/
Products
[93] T. Chalko. (2007) High accuracy speed measurement using gps (global positioning system).
[Online]. Available: http://nujournal.net/HighAccuracySpeed.pdf
[98] D. Bakker, “Battery electric vehicles - performance, CO2 emissions, lifecycle costs and ad-
vanced battery technology development,” Master thesis, Copernicus Institute University of
Utrecht, August 2010.
[99] D. Linden and T. Reddy, Handbook of Batteries, ser. McGraw-Hill handbooks. McGraw-Hill,
2001.
[100] A. Joseph and M. Shahidehpour, “Battery storage systems in electric power systems,” in
Power Engineering Society General Meeting, 2006. IEEE, 2006, pp. 8 pp.–.
204 BIBLIOGRAPHY
[101] M. G. Cugnet, M. Dubarry, and B. Y. Liaw, “Peukert’s law of a lead-acid battery simulated
by a mathematical model,” ECS Transactions, vol. 25, no. 35, pp. 223–233, 2010. [Online].
Available: http://ecst.ecsdl.org/content/25/35/223.abstract
[105] C. Ponce de León, “Energy conversion in fuel cells,” University Lecture. [Online]. Avail-
able: https://blackboard.soton.ac.uk/bbcswebdav/pid-2454138-dt-content-rid-1733817_
1/courses/FEEG6007-32779-14-15/Lecture%203%20Energy%20Conversion%20in%
20Fuel%20Cells%2014-19%20Nov%202014%281%29.pdf
[110] (2004, May) The Klien-Lindner axle articulation system. [Online]. Available: http:
//www.aqpl43.dsl.pipex.com/MUSEUM/LOCOLOCO/klienlindner/klienlindner.htm
[115] J. R. Calvert and R. A. Farrar, Elastic Bending and Torsion Relationships, 3rd ed., 2008, pp.
13–1.
[117] MatWeb. (2015) Matweb, your source for materials information. [Online]. Available:
http://www.matweb.com/
[118] P. M. Starink, Material Selection for Design, 2015, pp. 15–16. [Online]. Avail-
able: https://blackboard.soton.ac.uk/bbcswebdav/pid-2313206-dt-content-rid-1768721_1/
courses/SESG6042-32843-14-15/Ch1%20%28MJS2014%29%20compl.pdf
[122] The Chiel, “Diameter of train wheel,” 2013. [Online]. Available: https://answers.yahoo.com/
question/index?qid=20130620065628AANbLFa
[124] (2014) Relation between tractive force and motor torque. [Online]. Available:
http://www.researchgate.net/post/What_is_the_relation_between_tractive_force_pull_force_
and_motor_torque_in_traction_can_we_find_an_exact_relation_between_them_analytically
[125] M. Kozłowski, “Badania zerwania przyczepności w ruchu pojazdu szynowego (theory of rail-
way vehicles motion),” Prace Naukowe Politechniki Warszawskiej. Transport, vol. z. 74, pp.
3–104, 2010.
[128] Engineering Tool Box. (2015) Slope - degree, gradient and grade converter. [Online].
Available: http://www.engineeringtoolbox.com/slope-degrees-gradient-grade-d_1562.html
[129] Automotion Components, “Miniature linear rail carriages.” [Online]. Available: http://www.
automotioncomponents.co.uk/files/datasheet/L1010-miniature-linear-rail-carriages.pdf
[136] A. Bhaskar, “SESA6047 Aircraft Structures - Bending Analysis,” University Lecture, 2015.
[137] J. Astley, “SESM2016 Solids and Structures - Asymmetric Bending,” University Lecture,
2012.
[138] J. Astley, “SESM2016 Solids and Structures - Bending Shear Stress 1,” University Lecture,
2012.
[140] Wikipedia. (2015) Von Mises yield criterion. [Online]. Available: http://en.wikipedia.org/wiki/
Von_Mises_yield_criterion#Reduced_von_Mises_equation_for_different_stress_conditions
[145] D. Patrick, “Some notes on american locomotive practice 1948,” Journal of the Institution
of Locomotive Engineers, vol. 39, no. 207, pp. 54–86, 1949. [Online]. Available:
http://www.steamindex.com/jile/jile39.htm
[151] D. Thompson, “Sound transmition through partitions, isvr 3064 noise control engineering,”
University Lecture ISVR 3064, 2014.
BIBLIOGRAPHY 207
[152] C. Livingstone, “The assessment of noise pollution caused by bottle banks, and evaluation
of noise reduction methods,” University Assignment, ISVR noise control engineering (assign-
ment 2), 2015.
208
209
Silverline Bent External Circlip Pliers 230mm Tools £4.71 bearingboys BOM 595758 1 £5.65
Bearingboys Next-Day Delivery Bogies £8.50 bearingboys BOM 1 £8.50
Bosch Rexroth M8 Nut Bogies £7.53 RS BOM 390-0391 5 £45.18
M8x30 bolt Bogies £0.14 accuscrews BOM SEBF-M8-30-A2 32 £5.38
M5x90 bolt Bogies £0.58 accuscrews BOM SEBF-M5-90-A2 32 £22.27
M5 Nut Bogies £0.02 accuscrews BOM HPN-M5-A2 100 £2.40
M5 Wingnuts Bogies £0.12 accuscrews BOM HWN-M5-A2 32 £4.61
M8x25 bolt Bogies £0.13 accuscrews BOM SEBF-M8-25-A2 20 £3.12
M8x50 bolt Bogies £0.20 accuscrews BOM SEBF-M8-50-A2 20 £4.80
M8 Nut Bogies £0.05 accuscrews BOM HPN-M8-A2 100 £6.00
Mild Steel Plate (Dave H Drawbar) Frame £10.00 MetalSupermarket BOM 1 £12.00
Aluminium Plate 15mm (Charlie) Bogies £40.00 MetalSupermarket BOM 1 £48.00
Aluminium Plate 5mm (Charlie) Bogies £4.00 MetalSupermarket BOM 1 £4.80
Stainless Steel Plate 2mm (Charlie) Bogies £10.00 MetalSupermarket BOM 1 £12.00
Motor Mounts Fabrication (Mitch) Traction £10.00 Clarke-Lane BOM 2 £20.00
60mm M8 Bolts (Mitch) Bogies £1.55 Wickes BOM 3 £4.65
Brake Callipers (#1) Import Tax (Mitch) Brakes £43.29 ParcelForce BOM 1 £43.29
Cable Conduit, 25mm diameter, 10m long Electronics £34.81 RS BOM 513-2360 1 £41.77
Single Sided Strip Board Electronics £7.03 RS BOM 434-217 1 £8.44
ABS Enclosure Electronics £6.76 RS BOM 502-663 1 £8.11
Control Wire 100m, 7/0.2 BLUE, 1.4A Electronics £6.65 RS BOM 712-5336 1 £7.98
Control Wire 100m, 7/0.2 GREEN 1.4A Electronics £6.65 RS BOM 712-5333 1 £7.98
Control Wire 100m, 7/0.2 WHITE 1.4A Electronics £6.65 RS BOM 712-5355 1 £7.98
1A blade fuse (pack of 10) Electronics £1.54 RS BOM 787-4110 1 £1.85
2A blade fuse (pack of 10) Electronics £1.54 RS BOM 787-4126 1 £1.85
20A blade fuse (pack of 10) Electronics £1.54 RS BOM 787-4136 1 £1.85
5 x 3.9 Ohm Resistors Energy Rec £0.56 RS BOM 485-2580 5 £3.36
Screw connectors - ali frame Frame £9.53 RS BOM 420-6943 4 £45.74
6 POS Male Molex Electronics £0.44 Farnell BOM 2383183 15 £7.92
6 POS Female Molex Electronics £2.11 Farnell BOM 2383201 15 £37.98
Blade fuse and distribution Board Electronics £28.11 AES BOM 100134 2 £67.46
Delivery Electronics £3.75 AES BOM 1 £4.50
1/4 Turn Vacuum Valve Brakes £3.00 Easy Composites BOM VV-FF 1 £3.60
Delivery Brakes £1.30 Easy Composites BOM 1 £1.56
Clear PVC Tube - 3 1/16th Brakes £0.95 PNP Railways BOM PNR-1I 8 £7.60
Bosch Rexroth 45x90 2m Profile Frame £64.45 RS BOM 390-0133 1 £77.34
1.5mm Aluminium Sheet (EDMC) Shell £38.95 EDMC BOM 2 £77.90
1.5mm Aluminium Sheet (EDMC) Shell £29.50 EDMC BOM 1 £29.50
Delivery Shell £6.50 EDMC BOM 1 £6.50
Jockey Wheels Build Rig £15.84 Amazon 1OC 4 £63.36
Delivery Build Rig £11.44 Amazon 1OC 1 £11.44
3m Webbed Lifting Sling Build Rig £11.20 SafetyLiftingGear 1OC WEB3XLG 2 £22.40
Delivery Build Rig £4.99 SafetyLiftingGear 1OC 1 £5.99
Low Profile M8 Nuts Bogies £4.58 Precision Technology Supps BOM A43908 1 £5.50
Delivery Bogies £3.50 Precision Technology Supps BOM 1 £4.20
M8x80 bolt Bogies £0.28 Accuscrews BOM SEBF-M8-80-A2 38 £12.77
Delivery Bogies £4.95 Accuscrews BOM 1 £5.94
NI myRIO Panel Mount Kit Electronics £13.00 NI BOM 783091-01 1 £13.00
Delivery Electronics £14.06 NI BOM 1 £14.06
M6 Bobbin Energy Rec £3.38 Polymax BOM M6 60 ShA NR 4 £16.22
Delivery Energy Rec £4.50 Polymax BOM 1 £5.40
Plug + cable (Mitch) Traction £6.32 Portswood Hardware BOM 1 £6.32
Ratchet Straps (Mitch) Build Rig £5.99 eBay 1OC 1 £5.99
Nylon Sheet 500mmx300mmx10mm (Mitch) Bogies £25.82 RS BOM 282-0733 1 £30.98
Bosch Rexroth T-Head Bolt, 10mm (Mitch) Frame £8.56 RS BOM 390-0335 3 £30.82
Velcro Black Hook & Loop Strip, 20mm x 10m (Mitch) Frame £9.73 RS BOM 423-9555 2 £23.35
M8 Crimps (Mitch) Electronics £0.14 RS BOM 119-176 50 £8.40
M10 Crimps (Mitch) Electronics £0.35 RS BOM 795-1693 50 £21.00
50 Tooth Sprocket (Emergency!) (Mitch) Traction £17.87 eBay BOM 4 £71.48
10 Tooth Sprocket (Emergency!) (Mitch) Traction £1.52 bearingboys BOM PW108B/10 4 £7.30
Delivery (Mitch) Traction £8.50 bearingboys BOM 1 £10.20
D1400/0500 50mm Circlip (Pack of 10) (Mitch) Bogies £5.63 bearingboys BOM D1400/0500 3 £20.27
Delivery (Mitch) Bogies £8.50 bearingboys BOM 1 £10.20
Rubber Feet (Mitch) Electronics £8.39 Ebay BOM 1 £8.39
57mm Ali Tube (Harry) Traction £8.49 Ebay BOM 1 £8.49
19mm Ali Tube (Harry) Traction £6.25 Ebay BOM 1 £6.25
6mm Steel Keys Traction £6.09 Ebay BOM 1 £6.09
Loctite (Mitch) Frame £6.99 Halfords BOM 2 £13.98
Loctite (Harry) Frame £6.99 Halfords BOM 1 £6.99
Brake calliper mounts Brakes £20.00 Clarke-Lane BOM 1 £24.00
Brake interaction bracket Brakes £30.00 Clarke-Lane BOM 1 £36.00
Lockout plates Bogies £92.80 Clarke-Lane BOM 1 £111.36
Shelves Frame £136.80 Clarke-Lane BOM 1 £164.16
Sprocket adapters Traction £76.00 Clarke-Lane BOM 1 £91.20
B.
Assessment
Likelihood
improbable 1 1 2 3 4 5
Overall Likelihood
Overall Severity
Assembly/testing Magnetic Fields Magnetic fields can Operator Ensure a large visible warning is place over 1 4 4
211
interact with the braking system that electromagnetic
pacemakers/hearing fields are present.
aids
Risk Assessment
Assembly Magnetic fields Injury resulting from Operator Require anyone who works on the train to 3 2 6
magnetic jewlerry remove any metallic objects before they
interacting with work on the project.
magnetic fields.
Assembly//testing Harmful Fluid Puncture of brake Operator Protect the brake cylinder from puncture. 2 3 6
cylinder results in Ensure when brake fluid is being handled
contact with brake fluid. protective gloves are worn.
Assembly/Testing Pressure Implosion of vaccum Operator Ensure components installed that use 1 3 3
compents & potential vacuums are manufactured to an
hearing loss appropriate specification with safety factor
applied. Protect any depressurised
components from impact or puncture
through design layout. Use eye and ear
protection when maintaining braking
systems.
Page 1 of 4 HS/UOS/FR/038/04
212
Testing Brakes Brake failure results in Operator & Dedicated emergency braking system & 2 2 4
loss of control of the bystanders braking system will be designed to lock "on"
train. in event of failure
Testing Flywheel Flywheel or bearing Operator & Casing around flywheel sufficiently robust to 1 3 3
breaking under high bystanders contain flying fragments if event did occur.
load. Fragments and
shrapnel ejecting to the
surroundings.
Electrical assembly and testing Electrical Short circuiting of unit Operator & Wiring circuit to reduce risk of event to be 1 2 2
creating very high bystanders implemented.
discharge rates
exceeding safety
parameters of flywheel
or supercapacitor.
Electrical assembly and testing Electrical Hot surfaces to touch. Operator - Wires that are rated to withstand sufficiently 1 1 1
Burning up of wires equiptment high currents will be used.
caused by excessive damage
current in discharge of
supercapacitor or
electric flywheel.
Electrical assembly and testing Hazardous Fluids Overcharging / puncture Operator Rigidly maintained voltage input limits at 1 3 3
in Supercapacitor - risk of chemical burn below capacitor rated voltage to ensure that
due to voltage being run risk is minimised as much as possible.
above rated voltage of
the unit causing
generation of hydrogen
gas within the cell.
Puncture due to
increased pressure
within cell.
Assembly/testing Crushing Jack failure leading to Operator Jacks to have fail-safe measures 1 4 4
train falling implemented and train to feature jacking
points to secure the jacks in place to
prevent slippage. Working underneath to be
limited if practical.
Manufacture Welding Burns and damage to Operator & PPE to be worn and area to be clear, 2 2 4
eyes bystanders welding only to be carried out after
necessary training/inductions have occurred
APPENDIX B. RISK ASSESSMENT
Page 2 of 4 HS/UOS/FR/038/04
Testing Motor/Drivetrain Burns - the motor and Operator The motor and drivetrain will be adequately 2 2 4
drivetrain may become cooled and the operator will be told to avoid
hot in operation and contact with them whilst in operation. They
during testing. will also be reminded that components can
get hot by warning labels.
Testing Vehicle/Drivetrain Cuts/Amputaton/Wound Operator & Adequate safety guards will be put in place 1 4 4
ing - Drivetrain bystanders between any moving drivetrain components
components such as and locations that the operator or
chains, belts, gears and bystanders could be present in. Drivetrain
pulleys can break and components will be operated only within
when operating at high their stated limits with regard to forces
speed shrapnel can be transmitted and operating speeds.
produced.
Testing Electrical Electrocution - When Operator Terminals will be shielded and operator will 1 3 3 no
generator is in operation be instructed to not make contact with
it produces an electrical terminals whilst the generator is running.
current in its output
terminals.
Testing Noxious Gases Asphyxiation/poisoning - Operator & Generator is only to be operated outside or 1 3 3 no
When petrol or diesel bystanders in a correctly ventilated area with sufficient
generator is in exhaust gas extraction.
operation, it will emit
exhaust gases.
Testing Hazardous Fluids Burns - Liquid fuels Operator & Keep fluids secure in storage cupboard until 1 4 4 no
such as petrol or diesel bystanders needed. Enforce strict no naked-flame
are highly flammable policy when fuel is to be used. Correct filling
and could cause burns apparatus including funnel should be used
or injury if ignited. when locomotive fuel tank is to be filled.
Fire extinguisher should be on hand as
stipulated in general fire risk assessment in
this document.
Assembly Vehicle Tripping/falling over the Operator & Keep work area clear and tidy. 2 1 2
locomotive, bystanders
trailing/loose parts and
equipment.
Assembly/testing Electrical Electrocution and burns - Operator Keep all food and drink outside of the lab, 1 2 2
water/other liquids or well clear of the automotive.
getting in contact with
electronic equipment
and causing damage
213
Page 3 of 4 HS/UOS/FR/038/04
214
Assembly/testing Manual handling Personal injury resulting Operator Keep any spare/loose equipment locked 1 2 2
from someone in the lab away. Make it clear that the equipment that
getting hold of and is left out should not be tampered with by
tampering with the using signage.
controller or any other
loose equipment during
construction
Design of control systems on Desk-based RSI, back pain and eye Operator Adjust chair to right height at beginning of 2 1 2
LabView strain from poor posture session. Use of wrist supports. Take regular
and inadecuate breaks.
computer workstation
set-up.
Electrical assembly and testing Manual handling Personal injury resulting Operator Make sure all systems have no supply 3 2 6 no
from testing of control power when being installed. When testing
systems on the moving stay clear of locomitve. Design in fail-safe
parts of the locomotive methods of stopping all control systems.
e.g motors, energy
recovery system (if
using a flywheel)
Assembly/testing Fire Burns - Overheating, Operator & Ensure a fire extinguisher is at hand and a 1 3 3 no
wires shorting, bystanders saftey procedure is established. When
malfuctioning working on the train ensure power supplies
equiptment are disconnected and the train is grounded.
Testing Vehicle Crushing, bruising, Operator & Take extra care when moving the train or 1 2 2 no
impact injuries - moving bystanders bogies from a standstill, with enough space
train or bogies with between the train/bogies and any
others in immediate bystanders
vicinity
Electrical assembly and testing Electrical Electrocution and burns - Operator Using safe electrical practice during 1 3 3 no
wiring motors, batteries, electrical assembly and testing
etc. incorrectly or
unsafely
Heavy lifting Manual handling Back injuries, bruising Operator Using correct lifting techniques, getting 1 2 2 no
or crushing - lifting assistance where required or using the
heavy loads incorrectly crane if required.
or unsafely
Cutting Machinery Cuts, amputation - Operator Machine safety features and procedures 1 4 4 no
incorrect use of cutting resulting from required training having been
machinery carried out.
APPENDIX B. RISK ASSESSMENT
Page 4 of 4 HS/UOS/FR/038/04
C. Traction Data Sheets
Technical Data
Inductance @ 15kHz
Torque Constant
No Load Current
Armature Inertia
Speed Constant
Peak Efficiency
Resistance DC
Rated Voltage
Rated Current
Rated Torque
Rated Speed
Peak Current
Rated Power
Peak Power
Armature
Armature
Motor A Nm/A Rpm/V mΩ µH Kgm^2 kW % A kW Rpm V A Nm
126 18 0.055 140 185 8 0.0234 7 76 400 4.30 3360 24 240 12.2
LMC LMC
95 8 0.101 95 21 18 0.0238 16 88 400 7.2 3420 36 200 16.5
LMC LMC
LMC LMC
Marine Drive Syste
Label QTY PartNo. Description
A1 1 LEM 00010 P Shaft
LMC LMC
A2 1 LEM 00008 P Hub
A3 1 LEM 00009 P Cap
A4 2 LEM 00012 P Terminal post
A5 2 LEM 00011 P Terminal plate
A6 1 LEM 00004 P Band cover
A7 1 LEM 00006 P Commutator casting
A8 1 LEM 00007 P Drive casting
A9 1 LEM 00003 S Brush holder & Extrustions Sub-Assy
LMC
A10 1 LEM 00035 P Brush Cover
A11 16 LEM 00038 P Drive magnets
A12 1 LEM 00037 P Bearing
A13 1 LEM 00039 S Armature
Dimensions: MM
215
216 APPENDIX C. TRACTION DATA SHEETS
PRODUCT SPECIFICATIONS
ELECTRICAL BMOD0083 P048 B01 BMOD0165 P048 BXX
Rated Capacitance1 83 F 165 F
Minimum Capacitance, initial1 83 F 165 F
Maximum Capacitance, initial1 100 F 200 F
Maximum ESR DC, initial1 10 mΩ 6.3 mΩ
Test Current for Capacitance and ESR DC1 100 A 100 A
Rated Voltage 48 V 48 V
Absolute Maximum Voltage2 51 V 51 V
Absolute Maximum Current 1,150 A 1,900 A
Leakage Current at 25oC, maximum3 3.0 mA 5.2 mA
Maximum Series Voltage 750 V 750 V
Capacitance of Individual Cells11 1,500 F 3,000 F
Stored Energy, Individual Cell11 1.5 Wh 3.0 Wh
Number of Cells 18 18
TEMPERATURE
Operating Temperature (Cell Case Temperature)
Minimum -40oC -40oC
Maximum 65oC 65oC
Storage Temperature (Stored Uncharged)
Minimum -40oC -40oC
Maximum 70oC 70oC
*Results may vary. Additional terms and conditions, including the limited warranty, apply at the time of purchase.
See the warranty details and enclosed information for applicable operating and use requirements.
217
218 APPENDIX D. SUPERCAPACITOR DATA SHEETS
SAFETY
Short Circuit Current, typical
(Current possible with short circuit from rated voltage. Do not 4,800 A 7,600 A
use as an operating current.)
UL810a (B01 & B06 only,
Certifications RoHS
150 Volts)
High-Pot Capability12 2,500 VDC 2,500 VDC
TYPICAL CHARACTERISTICS
THERMAL CHARACTERISTICS BMOD0083 B01 BMOD0165 BXX
Thermal Resistance (Rca, All Cell Cases to
0.40oC/W 0.40oC/W
Ambient), typical8
Thermal Capacitance (Cth), typical 7,700 J/ Co
13,000 J/oC
Maximum Continuous Current (∆T = 15 °C) 8
61 A, RMS 77 A, RMS
Maximum Continuous Current (∆T = 40 °C)8 100 A, RMS 130 A, RMS
LIFE
DC Life at High Temperature1
(held continuously at Rated Voltage and Maximum Operating 1,500 hours 1,500 hours
Temperature)
Capacitance Change 20% 20%
(% decrease from minimum initial value)
ESR Change 100% 100%
(% increase from maximum initial value)
160%
Percentage change from value at 25°C
150%
140%
130%
120%
110%
100%
90%
80%
-60 -40 -20 0 20 40 60 80
Temperature (°C)
NOTES
1. Capacitance and ESRDC measured at 25°C using 8. ∆T=IRMS2 x ESR x Rca
specified test current per waveform below. 9. Cycle using specified test current per waveform
2. Absolute maximum voltage, non-repeated. Not to below.
exceed 1 second. 10. Cycle life varies depending upon application-specific
3. After 72 hours at rated voltage. Initial leakage current characteristics. Actual results will vary.
can be higher. 11. Per United Nations material classification UN3499, all
0.12V2 Maxwell ultracapacitors have less than 10 Wh capacity
4. Per IEC 62391-2, Pd = ESR x mass to meet the requirements of Special Provisions
DC
Dimensions
Unit: mm Dimension: 229(L) 138(W) 205(H) 211(TH)
Terminal T6
6 [0.63]
M6 [0.236]
6 [0.236]
138 2
211 2
205 2
180
229 2
0
Constant Current Discharge Characteristics : A (25 C ) Amps
F.V/Time 5min 10min 15min 20min 30min 45min 1h 2h 3h 4h 5h 6h 8h 10h 20h
1.85V/cell 94.1 74.0 62.9 52.6 41.8 31.7 25.9 16.5 13.1 10.7 8.60 7.49 6.08 5.19 2.83
1.80V/cell 126.4 94.6 76.0 62.2 49.4 36.8 29.0 18.0 14.1 11.4 9.23 8.03 6.45 5.50 2.86
1.75V/cell 142.5 103.9 83.1 66.9 51.3 38.2 30.4 18.7 14.3 11.6 9.46 8.25 6.56 5.56 2.89
1.70V/cell 156.9 113.3 88.7 70.4 53.3 39.7 31.3 19.4 14.7 11.9 9.71 8.42 6.65 5.61 2.94
1.65V/cell 173.0 122.2 94.3 74.7 56.3 40.7 32.4 20.0 15.3 12.4 9.98 8.61 6.75 5.73 2.98
1.60V/cell 190.8 132.7 100.8 79.6 59.4 42.5 33.6 20.7 15.8 12.7 10.3 8.79 6.82 5.79 3.00
221
222 APPENDIX E. BATTERY DATA SHEET
100
10 100C A
8 80
250C
6
Remaining Capacity(%)
B
5 60
Charging voltage :
4 0
30 C
2.25V/cell 400C
40
Life expectancy(year)
2 C
20
0
1 2 4 6 8 10 12
Storage Time(Months)
0.5
20 30 40 50 0
C No supplementary charge required
68 86 104 O A (Carry out supplementary charge before use if 100% capacity is required.)
122 F
Battery temperature Supplementary charge required before use.Optional charging way as below:
1.Charged for above 3 days at limted current 0.25CA and constant volatge 2.25V/cell.
B
2.Charged for above 20hours at limted current 0.25CA and constant volatge 2.45V/cell.
3.Charged for 8~10hours at limted current 0.05CA .
Avoid this storage period unless regular Top charge.
C Supplementary charge may often fail to recover the full capacity
12.0
0.25
120 15.0
110 Charged Volume
Terminal Voltage(V)
11.0
0.20
100
14.0 0.093C 0.05C
90 Charge Voltage 0.207C
80 10.0 0.4C
0.15 (Constant 2.25v/cell)
70 13.0
1C 0.628C
60 9.0
After 50% Discharge
50 0.10
12.0
40 After 100% Discharge
30 8.0
0.05 11.0 3C 2C
20
10 Charging Current
1 2 4 6 8 10 20 40 60 2 4 6 8 10 20
0 0
Min H
0 4 8 12 16 20 24 28 32 36 40
Constant Voltage -0.2Cx2h+2.4~2.45V/Cellx24h,Max. Current 0.3CA Temperature compensation for varying temperatures:
Constant Current 0.1C until the voltage reaching 14.4V,then 0.1Cx4h -Charge voltage -3mV/Cell/degC from 25degC norm.
45.0
R8.0 1.6 X 45 Chamfer
72
69
9
27.5
250
64.0 +.05
R20.0 -
.00
C C C
(10 MIN)
10
SEE DETAIL C
R8.0
(31.7)
5 HOLES OF DIAMETER 45mm
223
EQUISPACED AROUND WHEEL
EDMC Drawings
A 31.7
B 35.0 B
(R20.0)
(R32.0) SECTION A-A
R25.0 MAX
SCALE 1 : 3
NOTES:
- ALL SHARP EDGES TO BE CHAMFERED OR FILLET BY AN APPROPRIATE AMOUNT, UNLESS OTHERWISE
STATED, SUCH THAT ASSEMBLY P21-BOG-WHL CAN STILL BE ASSEMBLED.
- FINISH: FINE TURNED TREAD AND BORE.
- VALUES IN BRACKETS ARE REFERENCE VALUES AND SHOULD BE USED AS SUCH
DRAWN BY TOLERANCES UNLESS
Charlie Mehta OTHERWISE STATED
DO NOT SCALE LINEAR DIMENSION
MASS COST DESIGNED BY X = 0.5mm
X.X = 0.1mm
N/A Charlie Mehta
EDGES SHOULD ANGULAR DIMENSIONS
A X= 1 A
DATE SCALE XX = 1
BE FILLET 08/10/2014 1:2 ALL DIMENSIONS IN mm TITLE
A3 UNLESS OTHERWISE STATED
PROJECT CONTACT NUMBER MATERIAL TEXTURE SURFACE FINISH
10 MIN DETAIL C Wheel
GDP 21 07923377636 EN8
SHEET NO. OF ASSEMBLY NUMBER DRAWING NUMBER REVISION
SCALE 2 : 1 REMOVE ALL SHARP EDGES THE INFORMATION CONTAINED IN THIS DOCUMENT IS THE PROPERTY OF THE
UNIVERSITY OF SOUTHAMPTON. DO NOT COPY WITHOUT WRITTEN PERMISSION.
1 8 P21-BOG P21-BOG-WHL-001 1
IF IN DOUBT, ASK
6 5 4 3 2 1
224
8 7 6 5 4 3 2 1
(7.0 MIN)
(R25.0) (5.0)
F F
( 35.0) R2.0 MIN
(5.0)
IN
M
(R20.0 MAX) R1.0 MIN
.0
10
C
B
(R25.0 MAX)
10.0 MAX
E (1.25) E
DETAIL B DETAIL C DETAIL D
( 40.0)
SCALE 3 : 1 SCALE 3 : 1 SCALE 3 : 1
234.7
164.7
158.7
72.6
D D
59.4
FOR SLOT DIMENSIONS 46.2
PLEASE SEE SLOT
33.0
DIMENSION TABLE
19.8
6.6
D
5.0
40.0 MAX
35.0 MAX
50.0 MAX
C C C
5.8
85.8 6
52.4
99.0
SYMMETRICAL ON OTHER SIDE
113.7 OF LINE
SECTION A-A
A
4.75 INCHES MINIMUM
157.7
B B
NOTES:
- ALL SHARP EDGES TO BE CHAMFERED OR FILLET BY AN APPROPRIATE AMOUNT, UNLESS OTHERWISE
STATED, SUCH THAT ASSEMBLY P21-BOG-WHL CAN STILL BE ASSEMBLED.
- VALUES IN BRACKETS ARE REFERENCE VALUES AND SHOULD BE USED AS SUCH
SLOT DIMENSION TABLE - IF A CAD MODEL IS REQUIRED, I CAN SUPPLY IT. E-MAIL: CM14G11@SOTON.AC.UK
A (mm) B (mm) C (mm) - DIMENSIONS IN THE FORM XX.XX HAVE A TOLERANCE OF 0.01mm
35 33.0 MAX 1.6 MIN - GROOVES ARE FOR CIRCLIPS TO BE INSERTED, THUS CRITICAL TO GET RIGHT
40 37.5 MAX 2.0 MIN DRAWN BY TOLERANCES UNLESS
Charlie Mehta OTHERWISE STATED
50 47.0 MAX 2.2 MIN DO NOT SCALE LINEAR DIMENSION
X = 0.5mm
MASS COST DESIGNED BY
X.X = 0.1mm
N/A Charlie Mehta
ANGULAR DIMENSIONS
X= 1
A DATE SCALE XX = 1 A
01/12/2014 1:1 ALL DIMENSIONS IN mm TITLE
A2 UNLESS OTHERWISE STATED
PROJECT CONTACT NUMBER MATERIAL TEXTURE SURFACE FINISH
Axle
GDP 21 07923377636 EN8
SHEET NO. OF ASSEMBLY NUMBER DRAWING NUMBER REVISION
APPENDIX F. EDMC DRAWINGS
REMOVE ALL SHARP EDGES THE INFORMATION CONTAINED IN THIS DOCUMENT IS THE PROPERTY OF THE
UNIVERSITY OF SOUTHAMPTON. DO NOT COPY WITHOUT WRITTEN PERMISSION. 1 4 P21-BOG P21-BOG-WHL-001 1
IF IN DOUBT, ASK
8 7 6 5 4 3 2 1
6 5 4 3 2 1
10.0
A
D
DETAIL B D
SCALE 3 : 1
100
R1.20 Fillet
(R25.0 MIN)
R1.0 MAX
R1.0 MIN
10.0 MAX
B
C C
40.0 MIN
50.0 MIN
B B
R1.20
Fillet
(2.5)
200
A
5.0
SCALE 1 : 2
40.0 MIN SECTION A-A
DRAWN BY TOLERANCES UNLESS
Charlie Mehta OTHERWISE STATED
DO NOT SCALE LINEAR DIMENSION
MASS COST DESIGNED BY X = 0.5mm
X.X = 0.1mm
N/A David Wikramaratna
ANGULAR DIMENSIONS
A X= 1 A
DATE SCALE XX = 1
NOTES: 02/12/2014 1:1 ALL DIMENSIONS IN mm TITLE
A3 UNLESS OTHERWISE STATED
- ALL SHARP EDGES TO BE CHAMFERED OR FILLET BY AN APPROPRIATE AMOUNT, UNLESS
PROJECT CONTACT NUMBER MATERIAL TEXTURE SURFACE FINISH
OTHERWISE STATED, SUCH THAT ASSEMBLY P21-BOG-WHL CAN STILL BE ASSEMBLED. Brake Disc
GDP 21 07923377636 EN8
- VALUES IN BRACKETS ARE REFERENCE VALUES AND SHOULD BE USED AS SUCH SHEET NO. OF ASSEMBLY NUMBER DRAWING NUMBER REVISION
REMOVE ALL SHARP EDGES THE INFORMATION CONTAINED IN THIS DOCUMENT IS THE PROPERTY OF THE
UNIVERSITY OF SOUTHAMPTON. DO NOT COPY WITHOUT WRITTEN PERMISSION.
1 4 P21-BOG P21-BRK-001 1
IF IN DOUBT, ASK
225
6 5 4 3 2 1
226
6 5 4 3 2 1
25.0 127.0
D 21.0 D
R12.5 MIN R20.0 MIN
R8.5
15.0
50.0
17.0
A
C (4.0) C
R2.0
15.0
4.0 MIN
10.0
(21.0)
25.0
B B
R12.5
25
217
SCALE 3 : 4
OTHERWISE STATED. PROJECT CONTACT NUMBER MATERIAL TEXTURE SURFACE FINISH Suspension Adaptor
GDP 21 07923377636 Aluminum 6061
- VALUES IN BRACKETS ARE REFERENCE VALUES AND SHOULD BE USED AS SUCH.
SHEET NO. OF ASSEMBLY NUMBER DRAWING NUMBER REVISION
REMOVE ALL SHARP EDGES THE INFORMATION CONTAINED IN THIS DOCUMENT IS THE PROPERTY OF THE
UNIVERSITY OF SOUTHAMPTON. DO NOT COPY WITHOUT WRITTEN PERMISSION.
1 16 P21-BOG P21-BOG-SUS-002 1
IF IN DOUBT, ASK
APPENDIX F. EDMC DRAWINGS
6 5 4 3 2 1
1 2 3 4 5 6
G.
3rd ANGLE PROJECTION
A NOTE
BUSHES EITHER PRESSED or LOCTITED IN POSITION
WITH EQUAL PROTRUSION EITHER SIDE OF BAR
22 300 30
DRG.NO.
B
50
16 X 45°
(4 PLACES)
7/16" BORE SPH. BUSH
227
Drawbar Drawing
8.0
MATERIAL:- 50 X 8 MS BLACK FLAT TOLERANCES UNLESS STATED OTHERWISEISS.NO. AMENDMENT DATE:
NO DECIMALS - +/- 0,5mm
1 PLACE DECMLS. - +/- 0,1mm DRWN. BY: RC SCALE: N.T.S. DATE: 05/10/11
D FINISH:- 2 PLACES DECMLS. - +/- 0,05mm
THIS DRAWING IS CONFIDENTIAL & IS THE PROPERTY TITLE: DRAWBAR for STRAIN GAUGE TEST
SPECIAL NOTES:- OF THE F.S.M.R..
IT MUST NOT BE COPIED, LOANED OR SHOWN TO
THIRD PARTIES OR USED WITHOUT THE PRIOR WRITTEN
CONSENT OF THE F.S.M.R. DRG.NO. ISS.NO. 1
FRIENDS of the STAPLEFORD MINIATURE RAILWAY TEL: 01487-822-794 e-mail: richcol@talk21.com
A4
H. Bosch Beam Calculations
8981 500 201 11/08 Aluminum Framing Linear Motion and Assembly Technologies Bosch Rexroth Corp. 16-9
f
0.01 0.1 1 10 100 1000 f ( mm )
F
2
f
0.001 0.01 0.1 1 10 100 f ( mm )
F
3
f
F [N] I [cm4]
2000 000
10 000
90 x 360H
1000 000
50 60 x 60H 45 x 180H
5 000
4 000 40 45 x 60H
3 000 30 60 x 60 45 x 90H
20
45 x 60H
2 000
45 x 90
1 500 15 45 x 45H
45 x 45 22.5 x 180H
1 000 10
30 x 45
22.5 x 45
500 5
20 x 40S
400 30 x 30 1S 30 x 45
300 3 30 x 30 22.5 x 45
200 2 20 x 20
F
10 000
20 000
1000
1500
2 000
3 000
4 000
5000
6000
7 000
8 000
9 000
100
200
300
400
500
600
700
800
900
0
L [mm ] x y 16
x y
228
I. Additional Suspension Plots
229
230 APPENDIX I. ADDITIONAL SUSPENSION PLOTS
J. Bogie Curving Properties Code
from __future__ import division # because it makes sense
import math
import matplotlib . pyplot as plt
fl_wid = 0.00724
tr_wid = 0.026
231
232 APPENDIX J. BOGIE CURVING PROPERTIES CODE
p1 = rinner - disp2
p2 = whl2 [1] - fl_wid
# circles
circle1 = plt . Circle ((0 ,0) ,r , ls = ’ dotted ’ , fill = False ) # centerline
circle2 = plt . Circle ((0 ,0) , rinner , fill = False )
circle3 = plt . Circle ((0 ,0) , router , fill = False )
# rectangles
rect1 = plt . Rectangle (( whl1 [0] -0.5* chord , whl1 [1]) , chord , fl_wid , color = ’ black ’)
rect2 = plt . Rectangle (( whl2 [0] -0.5* chord , whl2 [1] - fl_wid ) , chord , fl_wid , color = ’
black ’)
rect3 = plt . Rectangle (( whl3 [0] -0.5* chord , whl3 [1]) , chord , fl_wid , color = ’ black ’)
rect4 = plt . Rectangle (( whl4 [0] -0.5* chord , whl4 [1] - fl_wid ) , chord , fl_wid , color = ’
black ’)
rect5 = plt . Rectangle (( whl1 [0] -0.5* tr_dia , whl1 [1]+ fl_wid ) , tr_dia , tr_wid , color = ’
black ’)
rect6 = plt . Rectangle (( whl2 [0] -0.5* tr_dia , whl2 [1] - fl_wid - tr_wid ) , tr_dia , tr_wid ,
color = ’ black ’)
rect7 = plt . Rectangle (( whl3 [0] -0.5* tr_dia , whl3 [1]+ fl_wid ) , tr_dia , tr_wid , color = ’
black ’)
rect8 = plt . Rectangle (( whl4 [0] -0.5* tr_dia , whl4 [1] - fl_wid - tr_wid ) , tr_dia , tr_wid ,
color = ’ black ’)
# axes
ax = plt . gca ()
ax . cla ()
ax . set_xlim ( -0.8* axldis ,0.8* axldis )
ax . set_ylim (0.98* r ,1.02* r )
ax . set_aspect ( ’ equal ’)
## # points
## ax . plot ( whl1 [0] , whl1 [1] , ’ o ’, color = ’ black ’)
## ax . plot ( whl2 [0] , whl2 [1] , ’ o ’, color = ’ black ’)
## ax . plot ( whl3 [0] , whl3 [1] , ’ o ’, color = ’ black ’)
## ax . plot ( whl4 [0] , whl4 [1] , ’ o ’, color = ’ black ’)
# lines
ax . plot ([ whl1 [0] , whl1 [0]] , [ whl1 [1]+ fl_wid , whl1 [1]+ fl_wid + tr_wid ] , ’ -k ’)
ax . plot ([ whl2 [0] , whl2 [0]] , [ whl2 [1] - fl_wid , whl2 [1] - fl_wid - tr_wid ] , ’ -k ’)
ax . plot ([ whl3 [0] , whl3 [0]] , [ whl3 [1]+ fl_wid , whl3 [1]+ fl_wid + tr_wid ] , ’ -k ’)
ax . plot ([ whl4 [0] , whl4 [0]] , [ whl4 [1] - fl_wid , whl4 [1] - fl_wid - tr_wid ] , ’ -k ’)
if showgraph ==1:
plt . show ()
return ( p2 - p1 ) *1000
def forloop () :
for i in range (0 ,1000) :
x = i /100.
a = curve (20 , x , showgraph =0)
print x , ’ meters ␣ = > ␣ ’ , a
return None