Ntnu 2012 PDF

Development and Construction of Vehicle
for Participation in the Shell

Eco-marathon Competition
Aslak Brage Espeland

Håkon Johan Seiness
Petter Thorrud Larsen
Hans Gudvangen
Product Design and Manufacturing

Submission date: June 2012
Supervisor: Knut Einar Aasland, IPM
Norwegian University of Science and Technology

Department of Engineering Design and Materials
PREFACE
This project would not have been possible without the help of our supervisors, sponsors, fami-
lies and friends. The team would like to sincerely thank all of the student who have helped us
during the project. We would specially like to thank:
Bjarne Stolpnessæter, Knut Einar Aasland, Børge Holen, Iver Johnsen and Per Øystein Nordtug at
the institute of product development and materials.
Eker Design and Paal Fediuk at High Performance Composites in Fredrikstad for helping us
produce the monocoque. Gylling Teknikk for making a second battery pack on very short notice,
just in time for the competition. SmartMotor for exceptional help with the motor controller. We
would like to thank the rest of the team for all the good times and great experiences.
Finally we would like to thank our main sponsor Det Norske Veritas (DNV) and Kristina Dahlberg
for their continued support of the NTNU Shell Eco-marathon team.
Trondhiem, June 2012
Aslak Brage Espeland Hans Gudvangen
Håkon Johan Seiness Petter Thorrud Larsen
ECO-MARATHON 2012 1
2 ECO-MARATHON 2012
ABSTRACT
A team of 13 NTNU students have developed and built a car to compete
in the Shell Eco-Marathon 2012 competition. This master project is a
continuation of the specialization project done in the autumn semes-
ter 2011. Production started in February and the car was ready on the
start line the 18th of May. Out of 22 competing teams in the battery-
electric category, the DNVFuelfigher2 came in 5th place. The best result
achieved was 163km/kWh which can be calculated into an equivalent
of 1581km/liter of gasoline. This report includes the design and produc-
tion process of all sub systems. In addition project management, system
engineering, media and the race itself is described.
SAMMENDRAG
Et team bestående av 13 NTNU-studenter har utviklet og bygget en
bil for å konkurrere i studentkonkurransen Shell Eco-Marathon Europe
2012. Dette master prosjektet er en videreføring av fordypningspros-
jekt gjennomført høstsemesteret 2011. Produksjonen startet i februar
og bilen sto klar på startstreken 18. mai. Ut av 22 konkurrerende lag i
batteriklassen, kom DNVFuelfigher2 på femte plass. Det beste resulta-
tet ble 163km/kWh som kan omregnes til en ekvivalent på 1581km/
liter bensin. Denne rapporten omfatter design og produksjonen av alle
delsystemer. I tillegg er prosjektledelse, system engineering, media og
selve løpet beskrevet.
ECO-MARATHON 2012 3
4 ECO-MARATHON 2012
ECO-MARATHON 2012 5
TABLE OF CONTENTS
1 INTRODUCTION 8 3.1.2.1 Boundary Conditions 72
3.1.2.2 Layup 73
1.1 Shell Eco-marathon 10 3.1.2.3 Resuslts 75
1.2 The Team 12 3.1.2.4 Conclusion 76
3.1.3 Designing the Molds 78
3.1.4 Production 80
2 NON-TECHNICAL 14
3.1.4.1 Milling 80
3.1.4.2 Layup 86
2.1 Project Management 16
3.1.4.3 Doors 92
2.1.1 Planning 18
3.1.4.4 Firewall 92
2.1.1.1 Individual Managment Plans 19
3.1.4.5 Wheel Well Covers 94
2.1.1.2 Creating WBS 22
3.1.4 Assembly 96
2.1.1.3 Scheduling Project 22
3.1.4.1 Monocoque 96
2.1.1.4 Budgeting 27
3.1.4.2 Wheel Well Covers 98
2.1.1.5 Risk Identification and Assessment 28
3.1.4.3 Doors 102
2.1.1.6 Risk Identification and Assessment
3.1.4.4 Surface Finish 104
Challenges 30
3.1.5 Windows 106
3.1.5.1 Molds 108
2.2 Systems Engineering 32
3.1.5.2 Vacuum Table 110
2.2.1 Knowledge Managment 34
3.2.5.3 Production 110
2.2.1.1 Capturing Knowledge with LEAN
3.1.5.4 Assembly 111
Techniques 36
3.1.6 Lights 112
2.2.1.2 Storing and Sharing Knowledge
3.1.6.1 Former Lights 112
with Model-Based SE 40
3.1.6.2 Interfaces 113
2.2.2 Visual Workflow Management 41
3.1.6.3 List of Requirements 113
2.2.2.1 Stand-Up Meetings and Visual
3.1.6.4 Materials and Technology 114
Project Board 41
3.1.6.5 Concept Development 115
2.2.2.2 Visualizing Project Progress 42
3.1.6.6 Final Concept 116
2.2.2.3 Avarege Risk Level as a Tool for
3.1.6.7 Production 117
Tracking Progress 42
3.1.7 Details 118
2.2.2.4 The Timeline 43
3.1.7.1 Door Details 119
2.2.2.5 Project Progress in a Micro-Level 44
3.1.7.2 Wiper 124
2.2.3 Verification, Validation and Testing
3.1.7.3 Back Hatch Mounting 129
Activities of the DNVFF2 44
3.1.7.4 Side Cover Mounting 130
2.2.3.1 Unit Testing
3.1.7.5 Towing Hook 132
2.2.3.2 Assembly Testing 44
3.1.8 Foil 134
2.2.3.3 Performance Test 46
3.1.8.1 Paint VS Foil 134
3.1.8.2 Inspiration 135
2.3 Promotion 48
3.1.8.2 Production and Mounting 138
3.1.9 Wind Tunnel Testing 140
2.4 Rotterdam 56
3.1.9.1 Wind Tunnel 141
2.4.1 Shell Eco-marathon 58
3.1.9.2 Experimental Setup 142
2.4.1.1 Track 58
3.1.9.3 Results 143
2.4.1.2 Technical Inspection 60
3.1.9.4 Conclusion 144
2.4.1.3 Testing 64
2.4.1.4 Race 65
3.2 Interior 146
3.2.1 Introduction 148
3 TECHNICAL 66 3.2.2 Dashboard 150
3.2.2.1 Last Cars Dashboard 150
3.1 Exterior 68 3.2.2.2 Interfaces 151
3.1.1 Introduction 70 3.2.2.3 Ergonomics 152
3.1.2 Strength Analysis 72 3.2.2.4 Analysis of Functions 153
6 ECO-MARATHON 2012
3.2.2.5 List of Requirements 154 3.4.1.2 Tie Rod Mount 236
3.2.2.6 Cooperation with Other Students 155 3.4.1.3 Cable Mount 237
3.2.2.7 Concept Development 156 3.4.1.4 Additonal Drag Link Parts 237
3.2.2.8 Final Concept 157 3.4.2 Steering Column 237
3.2.2.9 Placement of Buttons and Display 158 3.4.2.1 Analysis 238
3.2.3 Steering Wheel 160 3.4.3 Pulleys and Kevlar Cable 238
3.2.3.1 Last Cars Steering Wheel 160 3.4.4 Assembly and Testing 240
3.2.3.2 Necessary Functions 161 3.4.5 Race and Conclusion 241
3.2.3.3 Production Methods 162
3.2.3.4 Inspiration 163 3.5 Brakes 244
3.2.3.5 List of Requirements 164 3.5.1 Brake Pedal 247
3.2.3.6 Idea Generation 166 3.5.1.1 Pedal Arm 247
3.2.3.7 Ergonomic Testing 166 3.5.1.2 Brake Pedal Box 249
3.2.3.8 Three Different Concepts 168 3.5.1.3 Master Cylinders 250
3.2.3.9 Production 172 3.5.2 Other Brake Circuit Parts 250
3.2.4 Seat 174 3.5.3 Assembly and Testing 252
3.2.4.1 Former Seat 174 3.5.4 Race and Conclusion 253
3.2.4.2 Inspiration 175
3.2.4.3 Possible Materials 176 3.6 Wheels 254
3.2.4.4 Interfaces 177 3.6.1 Strenght Analysis 256
3.2.4.5 Basic Ergonomics 178 3.6.2 Production 257
3.2.4.6 List of Requirements 179 3.6.3 Performance 259
3.2.4.7 Idea Generation 180 3.6.4 Further Recommendations 259
3.2.4.8 Three Ideas 182
3.2.4.9 Selecting Ideas 185 3.7 Propulsion 260
3.2.4.10 Detailing of the Final Concept 186 3.7.1 Motor Design 262
3.2.4.11 Production 187 3.7.1.1 Requirements 262
3.2.5 Rear View Mirrors 188 3.7.1.2 Concept Description 262
3.2.5.1 Mirrors on DNVFF 188 3.7.1.3 Hubs and Axle 262
3.2.5.2 Inspiration 189 3.7.1.4 Rotors 265
3.2.5.3 List of Requirements 190 3.7.1.5 Rim Adapter 265
3.2.5.3 Idea Generation 191 3.7.1.6 Test Rig and SM-Adapter 266
3.2.5.4 Different Concepts 192 3.7.2 Motor Development 268
3.2.5.5 Production and Finished Product 193 3.7.2.1 Rotor 268
3.7.2.2 Stator Production 271
3.3 Suspension 194 3.7.2.3 Mold Production 271
3.3.1 Front Suspension 196 3.7.2.4 Open Casting 271
3.3.1.1 Hubs 204 3.7.2.5 Vacuum Casting 272
3.3.1.2 Axles 205 3.7.2.6 Spacer Ring 272
3.3.1.3 Steering Knuckles 206 3.7.2.7 Assembly of the Motor 273
3.3.1.4 Linkages 210 3.7.2.8 Problems Regarding the Construction 273
3.3.1.5 Mounting Points 214 3.7.2.9 Testing the Existing Motor 276
3.3.1.6 Conclusion 215 3.7.2.10 Conclusion 278
3.3.2 Rear Suspension 216
3.3.2.1 Introduction 216 2.8 Control System 280
3.3.2.2 Knuckle 216 2.8.1 Car Control System 282
3.3.2.3 Lower A-arm Connector 222 2.8.1.1 Characteristics 282
3.3.2.4 Coilover 224 2.8.1.2 Changes to the Car Control System 282
3.3.2.5 Toe Link Extension 227 2.8.2 Propulsional Control System 284
3.3.2.6 Rods and Rod Ends 228 2.8.2.1 Characteristics 284
3.3.2.7 Assembly 228 2.8.2.2 Changes in the Car Control System 284
3.3.2.8 Further Recommendations 230
4 CONCLUSION 286
3.4 Steering 232
3.4.1 Drag Link 235
3.4.1.1 Length Adjustment 235
5 REFERENCES 290
ECO-MARATHON 2012 7
1 IN
TRODUCTION
1.1 SHELL ECO-MARATHON
The Shell Eco-marathon (SEM) challenges high The vehicles may use one of the following fuel or
school and college student teams from around energy types:
the world to design, build and test energy ef-
ficient vehicles. With annual events in the Ameri- • Gasoline • Natural gas
cas, Europe and Asia, the winners are the teams
• Diesel • Hydrogen
that go the farthest distance using the least
amount of energy. • Biofuels • Solar
• Ethanol • Battery
The competition is devided into different vehicle
and energy categories: In 2012 the competition was held in an urban en-
vironment for the first time in Shell Eco-marathon
• Prototype: Vehicles in this category aim to Europe’s history. This was a way of showcasing
build the most aerodynamic and fuel-effi- the exciting student innovations and futuristic
cient vehicle possible. The rules for this class vehicles on a more realistic stage.
impose few limitations and allows for very Shell selected Rotterdam in the Netherlands
radical designs. to provide this urban stage given it is centrally
• UrbanConcept: The UrbanConcept category located in Europe. Shell Eco-marathon Europe
offers participants an opportunity to design 2012 ran from 17-19 May on the Ahoy track in
and build fuel-economy vehicles that are the middle of Rotterdam. The competition was
close in appearance to today’s passenger a huge success with many spectators and great
cars. racing happening on track.
10 ECO-MARATHON 2012
The story of NTNUs participation in the Shell Eco- the car a reality, supporting the team with
marathon started in the fall of 2007. During the materials, knowledge and production facilities.
first 4 years NTNU participated with the DNV Fuel The team designed, built, assembled and tested
Figher (DNVFF). This car used a hydrogen fuel cell the car in just 9 months, making it ready for the
as a energy source to power its electric motor. competition in May.
Many prizes and awards were achived with this
car, including two second prizes and a world When the car arrived in Rotterdam it was im-
record in its class. mediately recognized as an eye catcher. Several
teams, spectators and tv-crews wanted to have
The 2012 team decided to build a brand new a look at the car. The team was very happy to see
vehicle to continue competing in the Urban their hard work being appreciated by the crowd.
concept class, but this time using plug-in battery Due to extensive testing in Trondheim the car
technology. The new urban setting for the com- only needed minor adjustments before it was
petition demanded a more robust car and a well fully ready for the race. A total of three long
made suspension system. Experiences gained distance runs were completed; one test run and
throughout the years from the DNVFF meant that two race attempts (ny record for ntnu?). The
improvements could be made within every sub best result achieved was 163 km/kWh in a near
system. perfect run. The competition level in the battery
class was very high and the team ended up in 5th
DNV was willing to continue the role as main place.
sponsor for 2012. The car was appropriately
named the DNV Fuel Fighter 2 (DNVFF2) as a con-
tinuation of NTNU and DNVs cooperation in this
project. Transnova, a Norwegian governmental
organisation promoting environmentally friendly
transportation, sponsored the team with addi-
tional fundings.
Several other important sponsors helped make
ECO-MARATHON 2012 11
1.2 THE TEAM
The 2012 team consists of 13 master students from the
Norwegian University of Science and Technology
Fariborz A. Heidarloo
Industrial Engineering, Project Management, 2-year master
Project manager
Benjamin Gutjahr
Engineering Cybernetics, 5-year master
Control system
Fredrik V. Endresen
Energy and Environmental Engineering, 5-year
Motor
Aksel Qviller
Engineering Science & ICT, 5-year master Itsaso Yuguero Garmendia
Suspension Mechanical Engineering, 5-year master
System engineering
Petter Thorrud Larsen
Mechanical Engineering, 2-year master Hans Gudvangen
Exterior Mechanical Engineering, 2-year master
Suspension & rims
Aslak Brage Espeland
Mechanical Engineering, 2-year master Oluf Tonning
Exterior Engineering Science & ICT, 5-year master
System engineering
Mats Herding Solberg
Industrial design, 5-year master Eivind Sæter
Design Industrial design, 5-year master
Design
Silje Kristine Skogbrann
Media Communication and Information Technology, 5-year Håkon Johan Seiness
master Engineering Science & ICT, 5-year master
PR and media Steering and Brakes
2
NON-
TECHNICAL
2.1
PROJECT
MANAGEMENT
2.1 PROJECT MANAGMENT
Shell Eco-marathon 2012 was a New Product 2.1.1 PLANNING
Development (NPD) project. One of the distinct
characteristics of this type of projects is sig-
The planning stage comprises those course of
nificant uncertainty. According to Tatikonda &
actions executed to found the scope of the pro-
Rosenthal (2000) technology novelty and pro-
ject, clarify the goals, and develop the processes
ject complexity are sources of uncertainty while
needed to achieve those goals.
a team tries to make new product that can be
brand new or result of improvements on its prec-
According to Verganti (1997) two planning styles
edent. Comparing to regular NPD projects, Shell
may be utilized for new product development
Eco-marathon 2012 has unique feature which is
projects, feedback planning or reactive approach
not having marketing perspective. In high level,
and feed-forward planning or proactive ap-
the goal of project is learning through executing
proach. Choosing how the project is going to be
a project and challenge students’ knowledge in
planned in early phase in order to save time and
practical case. High level of uncertainty on the
money is another challenge. By using reactive
one hand and a project team that is formed by
approach, project team does not spend signifi-
non professional members (students) on the
cant time on solving uncertainties and gathering
other hand makes planning stage of the project
information for accomplishing tasks. This method
very prominent. Therefore, the focus of project
which is based on the notion of uncertainty
management chapter in this report is on plan-
reduction during project advancement, doesn’t
ning Shell Eco-marathon project.
need substantial financial resource and it is not
time consuming but it increases the risk of facing planning but if complete change is expected
late changes during execution phase that puts feed-forward planning was used.
sizeable financial burden on project. Unlike this, For accomplishing this stage through Shell
by following proactive approach, project team Eco-marathon 2012 project, following processes
shall allocate considerable time to think about were carried out: developing individual manage-
what may go wrong during project development ment plans, creating work break down structure,
and come up with solution for each opportuni- scheduling project, budgeting and risk identifica-
ties or constraints. Performing this method is tion and management.
expensive as processing data and having experts
for anticipating uncertain information need 2.1.1.1 Individual Management Plans
substantial financial support and the danger of Why previous leaders of NTNU’s Shell Eco-
losing track of time due to drowning into analyz- marathon team were not totally successful for
ing details. Project manager shall choose suitable handling all project management knowledge
technique for planning based on project char- areas? Why same level of management was not
acteristics. In Shell Eco-marathon 2012 project applied on various knowledge areas? The answer
both feedback and feed forward approaches for these questions can be inferred from project
were used. During initiating phase Table 2.1.1 reports of prior teams as following:
was prepared which illustrates the improvement
points in developing new vehicle based on study • Absence of management plans.
performed on status of various systems of prec- • Unsuccessful efforts to properly implement
edent car DNVFF. management plans during project lifecycle
This table helped project manager to know due to lack of experience.
which type of planning approach is suitable for
various parts or systems of the future vehicle. If Lack of experience is not an avoidable factor
the scope of change is not huge, then feedback since team members including project manager
are chosen among master students who are not plan of different systems of vehicle and project’s
experts. According to available project reports, goal needs to be defined during this time. A
preceding team leaders preferred to decide week after by taking team members’ individual
upon the way of managing different processes plans into account, team leader has to propose
right before their executing time. This approach an initial schedule. Project schedule should be
caused various troubles for team leaders such as realistic; meaning using large leads, lags or slack
losing track of time, inefficient data distribution is not approvable. Proposal may be reviewed and
and communication. Avoiding similar difficulties, modified by team during group sessions and
project manager needs to think about strategies be finalized (baseline) in one week. Shell Eco-
to manage and control different knowledge areas marathon 2012 will have deterministic weighted
(scope, time, cost, communication…) before schedule. The schedule shall contain work break-
starting with course of actions. Not only experi- down structure in two levels (excluding level
ence is an important factor to develop strategies 0), estimated duration of each work package,
(individual management plans), also stakehold- relations among them, their weight factor and list
ers’ opinion and comments, lessoned learned of important milestones. Preventing complexity
from preceding projects and university’s policies and delivering an easy to understand schedule,
are key factor to build a basis. Project manager resource allocation will not be considered in
shall not consider planning processes as one time schedule’s network calculations. For accurate and
task while request changes from stakeholders or realistic result out of project schedule. Updat-
project team, unexpected events and feedbacks ing work packages have to be weighted. Project
from executing and monitoring stages make manager is responsible for breaking the weight
reviewing and revising plans repetitively critical. factors and assigning them to project schedule
Precluding lack of enough time to proceed with elements. Microsoft Project 2012 is going to be
other project’s stages, it is important to spend used as time management tool. For tracking
sufficient time on planning processes. Therefore, time, each week team members have to send a
based on importance of project and required summary to project manager that includes ‘what
level of management, project manager needs to they have done’ and ‘what they will do’. Project
think how detailed the initial plans should be. schedule will be updated and be compared with
By taking progressive elaboration into account, baseline based on received summaries. Among
making a comprehensive and perfect plan on with different updating approaches available
inceptive steps which demands a lot of informa- in Microsoft Project 2012 software, recording
tion is misleading point of view about planning elapsed actual duration will be used. Results of
project. Following individual plans were gener- weekly evaluation have to be communicated
ated for Shell Eco-marathon 2012 project, after with authorized stakeholders. These weekly
few modifications: reports shall cover the status of project, roots
of variance and forecasts to be illustrated with
Time Management Plan tables, charts and especially S-curves. This plan
Team members have to read previous project can be altered by any change request confirmed
reports and choose their role in project within by project manager.
two weeks after introductory session. Next two
weeks should be assigned to think about master Cost Management Plan
Name of part Major Improvement Points Scope of Impact on Making

Change New Car
More complete shape and better aerodynamics,
Chassis Complete change High
Use of lighter materials, increase aerodynamics
Software: Two-way communication with the SMC,
Control System Complete change Mid
possibly using CAN-bus!
Engine Plates Weight reduction Slight change Low
Table 2.1.1: Improvement points for developing a new vehicle
Project team is responsible for developing list of engineer may define risks in different systems of
activities in order to estimate demanded financial the vehicle by involvement of other team mem-
resources. Proposed budget for securing spon- bers to prepare risk register. Project manager has
sors shall be calculated by bottom-up approach. responsibility to assure that not only technical
Rough estimations from team members have risks but also organizational risks are consid-
to be aggregated to calculate required funds ered. Preventing confusion, only qualitative risk
for higher work packages. Contingency reserve analysis will be performed to determine the likeli-
has to be considered in final approximation of hood and impact of each risk. With help of other
proposed budget. This reserved will be evaluated members, system engineer is in charge of prepar-
for each level 1 work package. If project team ing preventive or mitigation actions. Reviewing
is not successful to gather enough funds as as risk register has to be done each two weeks to
proposed budget, project manager is responsible ensure proper monitoring. All team members
to do top-down budgeting based on available are in charge of reporting emergence of new
money after securing sponsors. Project manager risk by weekly summary. If required, both project
has to have a categorization for different costs manager and system engineer have to make sure
within project to have better control over them. risk response is applied completely. Risk register
Spent amount of money on each system of may be updated anytime during project planning
vehicle needs to be communicated with project and executing by anyone in the team. In order to
manager by weekly summary. Project manager communicate status of project risk, online shared
is in charge of generating clear report of finan- spreadsheet on Google Document platform will
cial status of project that includes comparison be made to simplify data gathering and distri-
of actual expenses and baseline (budget) for bution. This plan can be altered by any change
each work package. Any purchase that values request confirmed by project manager.
less than 1500 NOK can be carried on without Although these plans were improved and de-
project manager confirmation. Project manager tailed by passing time but due to lack of enough
has authority to block further procurement and information at start-up to create comprehensive
money transaction of work package if associated framework, above mentioned plans sufficed
budget has reached its limit. As reviewing money to make a foundation for managing different
transactions and tracking expenses is essential, knowledge areas. Collecting individual plans
project team members must preserve procure- result in cohesive structure that is called “project
ment documents appropriately. management plan’” Project management plan is
endorsed, confirmed, live document that de-
Communication Management Plan fines project requirements, determine expected
Apart from any communication which includes outcome, and lead project execution and control.
technical data, team leader shall be the hub for This document is integral point of team leader’s
any external communication and responsible to responsibility and s/he needs to receive the
distribute information to right person in team. approval from key stakeholders and the commit-
Due to the fact that reviewing history of com- ment of team members on its content to solve
munications may be necessary any time in future, further conflicts as it becomes project control
project team has to avoid relying on exchanging reference.
information via telephone or any other methods
that they are not traceable. Using same email
address (e.g. NTNU’s) for any communication Individual Plans Challenges
is imperative. In case of sharing information By defining individual management plans, team
with whole team, individuals can use either leader determines what have to be done during
shared project calendar on Google or use pro- project life cycle. By this mean, she/he specifies a
ject mailing list. Precluding misunderstanding, framework which team members are expected to
interchangeable information must be clear and perform in. Apart from feature of the plans, strict
easy to understand. Language of all different or easy to deal, detailed or abridged, project
types of communication (verbal and non-verbal) manager shall not assume all of team members
should be English. This plan can be altered by any will accept management style right away. The
change request confirmed by project manager. point about Shell Eco-marathon project that
should not be neglected is how individuals form
Risk Management Plan the team, voluntarily. Therefore project manager
Project manager and system engineer are re- cannot impose own preferred style of manage-
sponsible persons to handle project risks. System ment plans to team members because it might
have negative impact on team performance or proper level of detailing in WBS. Level of detail
in the worst case may result in disbanding of one has influence on how project cost, duration and
or more members which put the project in huge technical complications will be taken care of. In
danger. Confusion, rejection or not having same another word, it deals with the manageability of
comprehension of project management plans, WBS segments. Reaching the smallest (lowest) el-
from team members was one of the challenges ements of WBS makes more information available
during planning stage of Shell Eco-marathon but it also has to be considered that it requires
2012 project. Due to facing rigid deadlines, re- more data processing and calculation. Sticking
sponsibilities and clear expectation from project to high level work packages makes the road of
manager, team felt limited and tried to bring achieving goal vague and less comprehensible.
comments in order to alter the content of plans Based on project goal, functional work break-
and make them easy to deal which mostly result down structure was chosen for Shell Eco-mar-
into less structured arrangement. Although some athon 2012 project. Different systems of final
opinions were considered project manager put deliverable (vehicle) were assumed as high level
endeavor to give the understanding that indi- work packages and almost the same logic was
vidual plans are required and essential to handle considered for decomposing them: specify-
project in integrate manner. If it is required one- ing reusable parts from previous vehicle if it is
on-one meeting with each team member should possible, making a list of general requirement
be held to ensure same level of comprehension is for the system, conceptual designing, finalizing
shared among them. Not following same path for design of system, production and/or purchasing
reaching goal or working out of project scope are of parts. Avoiding confusion, WBS was detailed
possible outcomes if team does not not work in down to two levels (excluding level 0) and sup-
predefined frame. Important rules and deadlines porting efforts such as PR and media activities or
shall be printed and put on the wall or a place team building sessions cannot be found in WBS
where everyone can easily see them as oral com- because this type of efforts did not have influ-
munication is not reliable method for conveying ence on scope of final deliverable but they are
important information that need to be last till considered in project schedule as milestones.
end of project. More levels were prepared by team members to
not miss any single task for delivering a system,
2.1.1.2 Creating Work Breakdown Structure but including them in work break down struc-
Work breakdown structure (WBS), (figure 2.1.1) ture was totally unnecessary as they could make
is a hierarchical arrangement which shows how time tracking, budgeting and resource planning
the project work is decomposed to manage- complicated. In presence of other team mem-
able and understandable pieces. This structure bers bottom-up assessment was done to assure
visualizes the project scope; meaning it defines all deliverables and tasks were involved. Tasks
what is and is not going to be done and delivered related to work breakdown structure of Shell Eco-
during project life cycle. As all of required steps marathon 2012 project was created as following.
to accomplish project has to be determined in Due to lack of information regarding time and
WBS, missing a work package may cause unfa- place of testing the vehicle as integrated whole,
vorable consequences. Depending on project this task is not mentioned in WBS and considered
manager’s preference and the purpose of execut- as milestone in project schedule.
ing project, WBS may be created in several ways.
Garcı´a-Fornieles et al. (2003) have introduced 2.1.1.3 Scheduling Project
following approaches (classifications) for mak- “Develop schedule is the process of analyzing
ing WBS: product oriented, process or functional, activity sequences, durations, resource require-
organizational, project life cycle, geographic loca- ments and schedule constraints to create project
tion of people. Suitable WBS is not only a tool for schedule” (PMI, 2008). Project schedule has differ-
understanding scope; also it is useful to estimate ent types of elements which may be used based
time and budget. Furthermore, work breakdown on scheduler’s choice, but following components
structure provides holistic view of project to are essential:
stakeholders which help them to comprehend
what project team is going to do. Jung and Woo • Activities: Lowest level of work breakdown
(2004) stated that the WBS provides a common structure also known as operation or tasks
view toward project for involved parties and it which are steps to accomplish work pack-
provides shared project language. Based on im- ages. Time, budget and resource can be
portance of project, manager has to decide upon allocated to this element.
• Milestones: A milestone is a principal ent that this method uses predefined probability
mostly indicates completion of a deliverable distribution (beta) that results into definite
or a major step during project execution. value, it is classified in deterministic ap-
• Precedence relations: This element shows in proach.
what sequence activities or milestones shall • Probabilistic (stochastic): in this approach, a
be to reach the goal. Four precedence rela- unique statistical distribution curve is as-
tions in duling are: Finish to Start (FS), Finish signed to each activity or work package
to Finish (FF), Start to start (SS) and Start to which determines the duration by generating
Finish (SF). a random number from defined curve.
Before start with scheduling, project Each approach has advantages and dis
manager needs to decide upon advantages. Deterministic approach is easy to do
the approach for generating the and understand. It needs less time and knowl-
framework. Generally, two approaches edge compared to probabilistic approach. But
exist: according to Pohl and Chapman (1987) determin-
istic scheduling is unable to take uncertainty into
• Deterministic: Assuming fixed duration for account and it leaves no opportunity for project
project schedule’s elements is fundamental manager to handle it. Although stochastic sched-
of this approach. Two methods are widely uling aims at considering uncertainty of activ-
use in this class: Critical Path Method (CPM) ity or work package, but this approach needs
and Program Evaluation and Review Tech- considerable experience and knowledge and it
nique (PERT). By using CPM, duration of task takes more time to make a framework compared
or work package will be a single number to deterministic one.
which is derived from past similar projects or
technical opinion of experts. In PERT, three Project manager has to put enough time to pre-
numbers, optimistic (minimum) time, most pare a flexible schedule that covers entire scope
likely time and pessimistic (maximum) time, of project, nothing more or less. Project schedule
are generated for estimating duration. As is a framework to include essential elements that
Figure 2.1.1: Work breakdown structure

without them the objective is not e.g. manpower. Nonrenewable resources without
attainable. This framework shall contain tasks or time restriction are accessible as long as project
work packages which estimating time, cost and runs e.g. project budget.
required resources for them is feasible. Sup-
porting tasks such as public relation and media Shell Eco-marathon 2012 had deterministic
activities is recommended to not be taken into schedule. CPM method was used to determine
account during scheduling because they are not work packages duration. This option was chosen
dependant on specific duration, and mostly not due to short available time for making sched-
schedulable. Activities with this attribute are ule and build easy to understand framework
called Level of Effort (LOE). Project schedule com- for involved parties. Estimating time was done
ponents have to have unique and clear name. for level two work packages. Based on detailed
This makes it more understandable for anyone breakdown that each team member possessed
and also it helps the project manager in time of for related system, approximate duration for
filtering or grouping segments. each work package was appraised. Summation
According to PMI (2008) three types of depend- of rough numbers from team members and
encies may be used during scheduling. Depend- reserved slack time for each work package was
encies determine precedence relations: considered as project duration. Project manager
should be aware that team members mostly try
• Mandatory: Mandatory dependencies are to assume the best condition in time of generat-
constraints which are forced by nature of ing values. So, it is up to him/her to think about
work. time safety margin for each component or whole
• Discretionary: Discretionary dependency, project to avoid lack of time because of and to
called soft logic, is restriction applied based have buffer in order to respond risks or deal with
on scheduler’s logic in order to handle ele- uncertainties. Milestones were used in two situ-
ments better. ations: For significant steps through project such
• External: External dependency is limitation as “body mold is available’” and when team had
from outside of project team such as stake- no control over task’s duration and it was totally
holders that might influence the schedule. dependent on external entity such as sponsor.
Figure 2.1.2 shows how milestones were ar-
In time of defining precedence relations, all ranged in schedule.
activities and milestones have to have succes-
sor and predecessor expect. First one which Based on experience and information from
does not need predecessor and last one which interviews with team members that provided
doesn’t need successor. So, it is logical to say that more technical insight, project manager defined
all activities and milestones should have at least the relations among work packages. All three
on FS or SS relation with predecessor(s) and at types of dependencies were considered during
least one FS or FF relation with its successor(s). It scheduling. The most influential dependency was
is the project manager’s responsibility to review external one from DNV (main sponsor) and Shell
all relations in order to ensure that none of the (competition holder) that imposed strict dead-
elements is dangled. Independent elements lines on project. SF relation was not use at all and
cause inaccurate tracking over schedule in time maximum duration for leads or lags was 4 days.
of updating it. Misusing leads and lags for com- Before proposing the schedule, all work packages
pressing or giving float without any strategy will and milestones were checked to not be dangled.
decrease the authenticity of the schedule as it To have accurate and realistic result from up-
will not show the real condition. grading the schedule, Shell Eco-marathon 2012
had weighted timing framework in which, each
Assigning resources makes schedule more pro- work package was given a value, according to
fessional but it should not be always an option its importance and duration. The total of weight
as it also makes the schedule more complicated factors shall be 1 or 100. Table 2.1.2 shows how
and hard to comprehend. Therefore, for prevent- weight was distributed among different elements
ing bafflement, if resource management is not of schedule. Important milestones has the most
important, resource allocation is better not to be weight, as very important elements such testing
done in project schedule. Kolisch and Padman the vehicle is included in it.
(2001) have defined two categories for project
resources: renewable and nonrenewable. Renew- Group session was held with entire team to
able resources are available within specific period review project schedule and ensure all relations
within a systems and among systems of the vehi- Sheduling Challenges
cle have correctly set. The outcome of this meet- How can the best fitted schedule for project be
ing was the baseline. S-curve related to baseline developed? Answering this question is a chal-
was sketched afterwards in order to be a tool to lenge for project manager during scheduling.
compare actual performance with planned one. Scheduling a project should be done by consid-
According to time management plan, Microsoft ering its resource, activities and performance
Project (MSP) 2012 software was chosen to im- measure characteristics. If project manager does
plement the framework. For sketching planned not put enough thought on identifying what is
S-curve which shows expected cumulative the status of three mentioned factors, she/he
progress, two new columns were defined in MSP might have a result that is not realistic. Herroe-
file, ‘Weight Factor’ and ‘Weighted %Complete’. len et al. (1997) have mentioned that for dealing
Following formula was set for Weighted %Com- with complexities during scheduling, knowing
plete column: its three features is essential. According to them,
three elements shall be assumed for resources to
[% Complete]*[Weight Factor]/100 analyze them properly:
By updating the project schedule weekly in 1. Number of resource type that can be zero,
software (moving status date), MSP calculates one or more than one.
%Complete column (cumulative progress) based 2. Whether the resources are renewable or non
on proportion of elapsed duration to total dura- renewable and their accessibility time period,
tion. But this does not count the importance if it is for whole project duration or specific
of element into account. That is why weighted one.
%Complete is more realistic value to rely on. 3. Availablity of renewable resources if it is in
Planned S-curve of Shell Eco-marathon 2012 constant amounts or in variable amounts.
project including weight factors was sketched as
shown in Figure 2.1.3. Six elements are recommended for understand-
ing the activties’ attributes:
As it is shown, no progress is expected between
weeks 14 to 17 due to Christmas break. The slope 1. If activities can be resumed in time of inter-
of S-curve is considerably increased between ruption or not.
weeks 24 to 30 because the work packages with 2. Constraints between activities that can be
huge weight factor had to be accomplished in mandatory, external or discretionary
this period. 3. The network of activities is probabilistic or
deterministic.
Figure 2.1.2: Milestones arranged in the schedule
4. Activities duration that can be random inte- Regarding performance measures, project man-
ger number, or random continuous duration ager shall know if penalty functions for delivering
or all tasks have same duration equal to Td. the final product, result or service in due time
5. Project deadline which means if there is no exist or not. Former condition is called regular
deadline, there is deadline imposed on activi- and latter is named non-regular measure. Mini-
ties and there is deadline on project. mizing the project delay is an example for regular
6. If cash flows are considered with activities or measures and maximizing quality of the project
not. Both amount and timing of cash flows is an instance for non-regular measure.
can be arbitrary or predetermined numbers.
Analysis of Shell Eco-marathon 2012 project
schedule by considering resources, activities
and performance measure characteristics are as
Name of Work Package Weight below:
Exterior 14
• Resources: project had two types of resourc-
Interior 9 es, work and material. Both renewable and
nonrenewable (with variable amount avail-
Propulsion 11
ability) resources were accessible for limited
Suspension 12 and specific period of time.
• Activities: Shell Eco-marathon 2012 project
Wheels and Rims 9
had deterministic schedule. Both resum-
Braking and Steering 9 able (testing a system) and non-resumable
(producing monocoque) activities were taken
Cybernetics 11
into account. Project manager thought over
Project Managment 8 three types of contraints in time of defining
relations among tasks. All of activities had
Important Milestones 27 integer number as duration which were not
Total 100 random as they were calculated based on
members’ judgement and historical data.
Table 2.1.2: Weighting of different elements Non of tasks had specified deadline but pro-
Figure 2.1.3: Planned S-curve of the project
ject had strict deadline of May 17, 2012 for manded money to finish project work: “Bottom-
competing in the race. Up” and “Top-Down’” In bottom-up approach,
• Performance measure: Schedule of project cost estimation has to be done for the lowest
were developed based regular measure, level components of work breakdown structure.
minimizing the project tardiness. Then estimation of activities which are related to
same work package shall be summed. By con-
2.1.1.4 Budgeting tinuing this, essential fund for the highest level of
According to Huang and Xu (1998) best solution WBS can be calculated. This approach has follow-
for financing projects with high uncertainty is ing advantages:
external subsidization. This is exactly how Shell
Eco-marathon project is financed every year. It • The most important advantage of this
is up to project team to find sponsors for project method according to Venkataraman and
and university plays no role in supporting project Pinto (2008), is forcing project team to make
financially. Kamien and Schwartz (1978) indicated detailed breakdown structure at planning
two main difficulties with external financing stage
compared to self-financing for project associated • Within this approach, as the knowledge of all
with high uncertainty: involved persons in project, even those who
are responsible for the lowest level activities
1. Finding external sponsor may be hard in such will be used for such important process; mo-
endeavors because if the project fails, few tivation of team members will be increased
tangible assets will be left which lender can due to feeling of involvement with project.
make claim on them.
2. Receiver of the fund may reveal some in- But this method also has a down side:
formation about project which might be
valuable for existing or new rivals of lender’s • By applying this method, the role of project
business. manager (or top management) in such criti-
cal task will be reduced at first steps as team
members in lower level of hierarchy who has
For attracting external sponsor(s), project team the responsibility of activities shall estimate
has to calculate proposed budget which will the costs and project manager has no control
be the basis of financial negotiations in further over the process till analysis are presented.
steps. • The chance of overestimating within bot-
An estimate should be calculated for all project tom-up approach is higher compared to
works or any element of project schedule that top-down because project team members
needs financial resource to be taken care of. This might tend to exaggerate the cost in order to
estimation that has to be approved by key stake- increase the flexibility.
holders is the budget. Taking all project costs • By considering before mentioned character-
into account is essential for budgeting. Project istics, repetitive adjustments to have best es-
manager has to classify costs in order to not miss timation is expected and this makes bottom-
any and have better control over them. There is up approach time consuming.
no single categorization logic for grouping them
and suitable approach may be chosen depend- In bottom-up approach, demanded financial
ing on project condition. resources is calculated and then provided but
in top-down approach, project team has to deal
As unexpected events which burden extra costs with predefined amount of money that has been
on project happen during execution, contingen- allocated to project by upper managers and try
cy reserve shall also be included in budgeting to proportion it appropriately among high level
process. Contingency reserve is critical financial work packages and this process continues to the
resources to reduce the risk of cost overrun. For lowest level of WBS. Top-down method has fol-
estimate the reserve, project manager needs to lowing advantages:
go through the result of qualitative or quantita-
tive analysis of risks. By considering the risks • “The advantage of top-down budgeting is
that have significant financial impact on budget, that top management’s estimate of project
required fund can be estimated. costs, in aggregate terms, often tend to be
quite accurate” (Venkataraman & Pinto, 2008).
There are two approaches for calculating de- • On contrary to bottom-up approach, project
manager has full control over disaggregat- what team planned for, budget was revised by
ing the budget to work packages, which top-down approach by project manager. Practic-
results in more accurate estimations and cost ing this method, available fund was distributed
control. among high level work packages which resulted
in following figure (cost baseline).
Disadvantages of this method are as following:
2.1.1.5 Risk Identification and Assessment
• Project costs might be underestimated due An undetermined event with either positive or
to cost saving. negative impact on project is called risk. One of
• Experience is essential to perform this meth- the critical responsibilities of project manager is
od. If project manager doesn’t have enough ensuring that project’s risk are identified and as-
knowledge or understanding about work sessed. Registering risks and their characteristics
packages and related activities, improper should be performed as first step. Involving pro-
fund allocation is largely possible. ject manager, project team and stakeholders in
this step is crucial. Team members’ participation
Shell Eco-marathon 2012 project costs were clas- is important as they have to take responsibility
sified into followings: of determined risks and their responses. Involve-
ment of stakeholders is critical as they provide
• Direct costs that are particularly related to supplemental goal information. Identifying risks
activities on project. Following items are and making associated plans to respond them
included in this category: for whole project life cycle at planning stage is
• Human resource costs: this includes salary not possible as new risks emerge during project
of two team members that had to be paid advancement. Because of this, project manage-
because of their student assistant con- ment team should analyze the status of project
tract with department. iteratively in order to be sure that all kinds of risks
• Material Costs are considered. Project feature (low/high uncer-
• Operational Costs tainty) defines how often such analysis shall be
• Indirect costs which do not have direct im- done. According to PMBOK (2008) five processes
pact on the work of project. Mostly, adminis- can be accomplished in planning stage to follow
trative expenses are put in this category. best practices in risk management area:
Budgeting this project requires both bottom-up 1. Plan risk management: Within this process,
and to-down budgeting approaches. First for project manager define how risks are go-
negotiating with sponsors, project team needed ing to be handled through project life cycle.
to come up with proposed budget. In this order Importance of making this plan is having
based on historical data from previous project an accepted basis among team members to
reports and team members’ judgments, rough evaluate undetermined events. Risk manage-
calculations were done for WBS work pack- ment plan is explained in section.
ages. Using bottom-up approach, the proposed
budget was estimated at 1.095.000 NOK includ- 2. Identify risks: Result of finishing this process
ing contingency reserve. Table 2.1.3 shows more is a list of categorized risks. Chapman (2002)
detail about it. has stated that although present models
and methods in project management make
Lots of effort were put by project team and man- valuable tools available but still experts’
ager to secure sponsors in order to decrease the judgments are the key input for identifying
amount of demanded money in different items project risks. This process is fundamental for
of the budget. Beside, by gaining more informa- all of risk management efforts; therefore ac-
tion about structure and design of various sys- curacy is very important. Various methods are
tems, team found overestimations in some items available for gathering such inputs. Chapman
which was expected after utilizing bottom-up R. J. (1998) has classified all methods into
approach because for not facing any trouble in- three groups:
dividual’s assessments were associated with high • Identification managed by one risks ana-
uncertainty. At last, NTNU’s team could receive lyst e.g. reconsidering historical data.
financial contribution from Det Norske Veritas • Identification performed by analyst inter-
(DNV) of 600.000 (main sponsor) and Transnova viewing project’s key players.
of 150.000 NOK. As taken fund was less than • Identification conducted by presence of
Budget item Planned Contingency all key players in form of group sessions
Cost (NOK) Reserve that is leaded by analyst e.g. brainstorm-
(NOK) ing.
From above mentioned classes, first and
Propulsion 150.000 20.000 third were used in Shell Eco-marathon 2012
Exterior 300.000 50.000 project. First, different engineers based on
historical data from previous years’ reports
Wheels and Rims 55.000 5.000 and their experience, tried to list risks associ-
Suspension 110.000 20.000 ated to systems that they were responsible
for. Afterwards, by using scenario building
Braking and Steering 15.000 5.000 (analysis) technique in group sessions leaded
Interior 10.000 5,000 by project engineer and supervised by pro-
ject manager, risks related to system interfac-
Cybernetics 50.000 10.000 es were determined. By using this technique
various possible events and their outcomes
Shipments 40.000 10.000
were identified. While scenario building, one
Trip to Rotterdam 142.000 8.000 does not count historical data and does not
include them into assessment; but it provides
Misc 40.000 0
a picture of future which is linked to past. Be-
Safety Margin 50.000 side technical (low level) risks, team members
and project manager identified managerial
Total 912.000 183.000
and organizational (high level) risks which are
mostly connected to team, its performance
Table 2.1.3: Detailed budget and managing project.
Figure 2.1.4: Detailed budget
the likelihood of risk occurrence and its
3. Perform qualitative risk analysis: The purpose impact that mostly measured in terms of cur-
of following this process is organizing of rency. According to Apostolakis (2004) this
identified risks in order to enhance execution approach has following benefits:
of project by concentrating on high-priority • Delivering thorough comprehension of
risks. High-priority risks are those which have system failure manners by considering
significant impact on time, cost and quality significant number of scenarios include
of project. For deciding how critical each risk different style of failures.
is, two criterions are used: probability and • It is unified method, so determining the
impact. For assessing subjectively, a scale requirements from various disciplines
shall be defined for each parameter in order involved in project.
to have common understanding of rating • By using this approach, the chance of
system. Depending on project manager and taking intricate interactions between
project feature defined structure for scales systems and operators into account will
varies. Often, “High, Medium, Low” or nu- be increased.
meric “1-10’”scale is used for measuring both • Unlike qualitative risk assessment ap-
probability and impact of risk, if occurred. proach, output of quantitative method
Project manager needs to be aware that in can be used in cost-benefit analysis.
qualitative assessment “risk is relative to the
observer” (Kaplan & Garrick, 1981). Specified This technique has following limitations:
risk characteristics may differ from one team • Not modeling human errors in time of
member to another as the knowledge and facing risk
experience of the observer are important • Not considering the culture of people
factors in organizing risks. Probability and im- who have the responsibility of handling
pact matrix was used in order to accomplish the risk, as it is influential factor in how
qualitative risk analysis in Shell Eco-marathon individuals will react when accidents hap-
2012 project. Numeric “1-5’”rating system was pen
fixed for assessing two parameters of each • Not taking design and production errors
risk. Numbers have following definition given into account.
in table 2.1.4. This approach wasn’t used in Shell Eco-mar-
athon 2012 project because it is time con-
This tool helps project team to know which suming and team members didn’t have the
risks need immediate response and which experience and knowledge about following
ones needs to be reviewed later as they are quantitative risk analysis.
not critical. Qualitative risk analysis has some
advantages and disadvantages. Startienė and 5. Plan risk responses: Considering the outputs
Remeikienė (2007) have indicated followings of identify risks process and all possible up-
as its advantages: coming and expected events, team members
• It is useful when enough experience is developed mitigation actions individually (for
not avaliable. responsible system) or in group (for inter-
• It is flexible as scales can be altered easily faces of systems). Also, project manager build
because no complicated calculations exist a list of responses for managerial and organi-
behind them. zational (high level) risks.
• It is less time consuming and cheaper
compared to quantitative risk assessment 2.1.1.6 Risk Identification and Assessment
Challenges
But this approach has following downsides There is no guarantee for reaching expected out-
as well: come out of group sessions that are dedicated to
• Less precise compared to quantitative risk identification and assessment. What makes
approach because results are shown in it uncertain is the group effectiveness during
subjectvely manner. meetings. This is a challenge for project manager
• Cost-benefit analysis cannot be done with to comprehend the status of team’s effective-
outputs from this method ness and make proper decision if it is low. This
evaluation is very critical because in time of low
4. Perform quantitative risk analysis: For accom- efficiency, not only time is wasted but also the
plishing this process, team tries to quantify results are not reliable and the chance of facing
troubles during project advancement will be in- • The environment: It is important that
creased if risk management processes are based participants in risk identification and as-
on untrustworthy judgments. Chapman (1998) sessment meeting feel that the location
has developed a model for evaluating group ef- of session is proper. Quiet room that has
ficiency. According to model two factors can be enough facilities where people can find
assessed related to group sessions: “The Givens” a comfortable seat gives good sense to
and “Intervening Factors”. In his paper each factor team member and they can carry on the
is detailed into few determinants but in this re- tasks in effective way. Members don’t
port only those that are connected to the project want to spend their time in assemblies in
are discussed. which the outcome of it is not communi-
cated or put into effect. Therefore, team
• Givens: this feature describes the status of leader or director of meeting should pro-
the group, the tasks and the environment vide an environment in that participants
that are inputs for risk identification and as- feel its significance for project e.g. to hold
sessment. the meeting in structured or organized
• The group: Size of the group matters. manner.
Although increasing the number of the • Intervening Factors: The most important item
group for each session guarantees the in intervening factor is motivation. Risk iden-
involvement of various disciplines and tification and assessment sessions should be
knowledge but project manager should presented by team leader in way that team
be aware that this might result in decreas- members wish to be part of them and ap-
ing the individual contribution. Compat- prove the objective of meetings.
ible members shall be present in meet-
ings. Discordant members will decrease For Shell Eco-marathon 2012 project, risk iden-
the effectiveness of group work. While tification was done by individuals and in group
productivity of sessions is the common for ensuring, each team member benefits of own
goal for all participants, project manager contribution and team work. Meetings weren’t
shall have this awareness that members held for more than five members including sys-
may try to include their personal objec- tem engineer as director. Director had the knowl-
tives as well e.g. imposing own interest edge about technical group gathering to not let
or trying to grab the lead. Therefore, an the members distract the flow of proceedings
unbiased person who has enough knowl- by their personal interest. All of team assemblies
edge to guide the meeting should be were carried on in presentation room in order
present. to provide appropriate physical location. Before
• The tasks: Team leader must be sure that each meeting, participants were informed about
team members take their responsibilities agenda and topic, so they could make their opin-
seriously and handling various tasks is im- ions ready. Minute of each meeting was prepared
portant for them. Individuals show more of system engineer to communicate the result
commitment if they consider the task with all team members with purpose of show-
prominent. Lucidity of the tasks is impor- ing the importance of assemblies to them. Not
tant factor on group effectiveness as well. only project manager explained the significance
When the expected performance and of this step to team members, but also director
outcome is less ambiguous for partici- of meetings, clarified why project needed it and
pants in a meeting who have same level what the goal was. Therefore, everyone had clear
of comprehension about what they have and common understanding of process and its
to do, effectiveness will be increased. objective.
Scale Probability of Occurrence/Impact Related Color
1 Very Low
2 Low
3 Medium
4 High
5 Very High
Table 2.1.4: Definition of rating system
2.2
SYSTEMS
ENGINEERING
2.2 SYSTEMS ENGINEERING
The SEM 2012 team has integrated for the first 2.2.1 KNOWLEDGE MANAGEMENT
time a systems engineering team. The Systems
Engineering team consist of two system en- One of the Systems Engineering tasks of 2012
gineers. The SE effort lays on the knowledge was to secure a seamless transition from this
management, Visual Workflow Management, risk year’s team to the next, and to put in place a
analysis and management and in designing and system for capturing and sharing knowledge for
performing a technology qualification. years to come. This is the topic for Oluf Tonning’s
master’s thesis, and more details about this will
be found there (Tonning, 2012). This section gives
a summary of the contents of Tonning’s thesis.
Figure 2.2.1: The template uses a Plan-Do-Check-Act approach, meaning the problem is solved
through analysis, testing, verification and then implemented
2.2.1.1 Capturing Knowledge with LEAN K-brief that Tonning (2012) has designed for cap-
Techniques turing knowledge about sub-systems and com-
A much-employed technique in Lean Thinking ponents of the DNVFF2. This technical K-brief is
(a theory for efficient development and produc- meant as a deliverable at the end of the process,
tion from Toyota) is the Knowledge brief, better summarizing the entire development, produc-
known as K-briefs. These are A3 sheets of paper tion and utilization process.
containing short and visual explanations of a
lesson learnt or a problem encountered and An example of a K-brief are available in Appendix
solved. The idea is to make the K-brief inviting to G.
the reader, and to explain to the reader in short,
summarized terms how to deal with a problem
or about the decisions leading up to a certain
design choice made in the development process.
To best capture knowledge, the process needs
to be standardized. Lean Thinking recommends
having certain learning events at equal intervals
during a development process. The most com-
mon is to have many milestones throughout the
development process, and to have the learning
events at these milestones. The teams will spend
that day or period capturing knowledge from
that interval. In particular interest are problems
solved and decisions made.
Another way of standardizing the process is to

have specific templates for the K-briefs. Tem-
plates reduce the time spent while making these
K-briefs, as well as securing that nothing is left
out that should have been included.
Figure 2.2.1 shows a template for a K-brief that

addresses a problem solved.
The following three sides show a three-page
Figure 2.2.2: The front page is dedicated to listing components, interface diagrams, describing major decisions
and the manufacturing process. The components are categorized as ‘New Product Development’ (NPD) meaning
designed by this year’s team, ‘Reused’ (R) meaning an unaltered design or ‘Purchased’ (P) meaning bought from a
manufacturer. Also, the components are rated for satisfaction stating how well the component behaved relative to
its expected behaviour ECO-MARATHON 2012 37
Figure 2.2.3: Page 2 is dedicated to the 3D-models, simulations and analyzes
Figure 2.2.4: Page 3 is dedicated to perceived risks, a description of problems and how they were solved, and
proposed future work
Figure 2.2.5: The sub-system “Wheel” broken down into its components
2.2.1.2 Storing and Sharing Knowledge with

Model-Based Systems Engineering
Now that the team has decided on how, when
and what to capture, the team needs to have a
“where” to put it. Model-based Systems Engi-
neering (MBSE) is a way of performing Systems
Engineering through system modelling. A system
model is a three-step description of a system,
describing it by its requirements, functions and
system architecture. The model uses hierarchies
to illustrate how elements within model are relat-
ed to each other, and to add detail to elements.
Figure 2.2.5 shows how the element Wheel can
be further described by breaking it into its minor
elements.
Figure 2.2.6 shows how an N2 diagram may be

utilized to show how the components of the
Wheel are fixed together, or i.e. the interfaces
between components.
Requirements, functions and components are Figure 2.2.7: A requirement is satisfied by a function
strongly related to each other. A requirement and assigned to a component. A function satisfies
says why a function or component is chosen. A a requirement and is realized by a component. A
function describes what a component or sub- component realizes a function and is associated
system is intended to do, and components are with a requirement
Figure 2.2.6: The elements that make up the wheel are positioned along the diagonal. Lines between the elements
show interfaces between them (emphasized by diamonds). The Hub is the part that links the sub-system to the rest
of the system
the actual realization of the concept idea. Figure 2.2.2 VISUAL WORKFLOW MANAGE-
2.2.7 explains the relationships.
MENT
It is imperative that the system model show these
relationships. When studying a single element in The Systems Engineering team employed Visual
the model, the model must visualize the relation- Workflow Management (VWM) to monitor pro-
ships this element has with the rest of the model. ject progress, enhance communication between
team members and spot problems before they
The decision to use MBSE in knowledge manage- occur. The tools used are Stand-up meetings
ment is based on MBSE’s perceived benefit of with a tailored Visual Project Board, risk man-
increasing learning and easier training of new agement, and two methods developed by the
members of a team. This is achieved by so-called SE team themselves. Details about these tools
Model-based Documentation. This means linking may be found in Oluf Tonning’s master’s thesis.
documentation directly to the element in the This report gives a summary of the tools and the
model of which is describes. E.g. a K-brief about process.
the production of the windshield could be linked
directly to the element Windshield in the model. 2.2.2.1 Stand-Up Meetings and Visual Project
This way, the model acts as a repository for Board
information. Anyone searching for information Stand-up meetings are short team meetings
about a certain element may navigate through where the team communicated to each other
the model to the element-of-interest and retrieve what they have done since last time, what will be
a list of documents, CAD files or photos that done till next time and whether or not they are
directly describe the element. For PDF files, the facing problems.
hyperlinks may also link directly to specific pages
within the document. Tonning tailored a Visual Project Board (VPB) for
using during the meetings. This is a whiteboard
The system model is found in the root directory template, where every team member lists their
of the server space for the 2012 team. It opens in tasks over the next fourteen days. Colour codes
the web browser, and the user navigates through tell whether a task is on-schedule or slipping.
the model much like Wikipedia. Figure 2.2.8 shows the VPB for DNVFF2.
Figure 2.2.8: VPB for DNVFF2. Tasks are numbered on the right-hand side. The Gantt-diagram in the middle shows
what day a certain task is due. Colors tell whether a task has slipped or not. The left-hand side is for information and
exchanging messages
The SE team made these conclusions about • Average risk level; for early discovery of prob-
Stand-up meetings and the VPB: lems and mitigation
• Timeline; for overall system progress and om-
• Maximum number of active participants munication
is 7 or 8 • Wall-Architecture; for system progress on a
• 12 minutes is the maximum time micro level and communication
limit.
• Location matters during production periods. 2.2.2.3 Avarege Risk Level as a Tool for
The meetings must be held in or Tracking Progress
close to the workshop. The system engineers made a huge effort in the
risk analysis and management of the project. The
• Addressing the three questions (“what have first step in the risk management process was
I done since last time”, “what will I do till next to identify the risk. To do so a “risk session” was
time” and “what is preventing me from pro- organized. By arranging “risk sessions” the team
gressing”) makes the meetings shorter and identified and analyzed risks on a component
more relevant to the listeners level. Using a 1 to 5 scale, the team assessed the
• The meetings are best held in the morning, likelihood and impact of each risk.
during production periods due to the acces- The risks were stored in a Google spreadsheet
sibility of machines. that everyone had access to. The average risk of
• Stand-up meetings are very important during
the competition, due to the ever-changing
environment that a competition is.
• Stand-up frequency
• During production: three times a
week is sufficient
• During competition: At least twice a day
• VPBs are great tools for managing team
work on a micro-level, and for communi-
cating important messages between team
members.
2.2.2.2 Visualizing Project Progress

The SE team used three tools for tracking and Figure 2.2.9: Risk cube display
visualizing project progress. The argument
for using three different methods was to gain the system was shown as a curve, updated on a
alternative perspectives on the project. The three weekly basis, as shown in figure 2.2.10.
methods are:
Figure 2.2.10: Curve showing how average risk level developed through the project
This way of visualizing project progress is a very early as possible. Since the SEM 2012 team deliv-
powerful tool for identifying possible problems, ers their risk analysis to the next year’s team – an
and monitoring whether the team is good at analysis based on experience and knowledge –
solving problems as they are identified. the next year’s team may utilize this tool to its full
In industry related projects the amount of risks potential.
that must be still in the project when the produc-
tion phase is entered should be as low as pos- 2.2.2.4 The Timeline
sible. For monitoring overall project progress the SE
team developed a way of visualizing every sub-
system’s progress using a timeline. This timeline
was drawn on the big whiteboard in the office,
next to where the team has its Monday meetings.
The timeline was also meant as a place to show
important happenings, and to induce a competi-
tive attitude in the team.
The Timeline divides the project into one-week

Takt periods. The responsible team member for a
sub-system listed the major tasks for every week
for the rest of the period of that sub-system. The
sub-systems were indicated on the Timeline with
pieces, like in a board game. The piece would
Figure 2.2.11: Variation of program risk and effort move closer to the finish line (competition date)
throughout system development (Kosiakoff et al, relative to the amount of tasks finished. E.g. a
2011) sub-system finishing two thirds of its tasks over
three weeks will only move two weeks on the
However, for this project it did not do as well. Timeline. The Timeline was updated on the Mon-
Due to lack of expertise and time most of the day meetings.
risks were identified in the production phase. The
team was not mature enough when it comes to Team members who were not native to the office,
working with risks, and there was not enough found the Timeline useful for gaining informa-
time get the experience needed or enough time tion on happenings and progress. Critique of the
to finish a proper analysis. More and more risks Timeline state it should have been employed
were identified, only adding to the risk level. earlier, with a stronger focus on work planning.
The SE team decided to stop using this tool for Overall, the Timeline is a very useful tool that may
visualizing project progress to avoid unnecessary gain much popularity within a team.
stress in the working group. However, the team
sees that this is a very useful tool if it is applied as
Figure 2.2.12: Template for the Timeline

2.2.2.5 Project Progress on a Micro-Level; The phases; Unit testing, Assembly test, Performance
Wall-Architecture test and Race test.
To monitor progress on a micro-level, the SE More detail about the testing strategy can be
team made use of the system architecture found at I.Yuguero-Garmendia (2012).
made for the system model (in the knowledge
management). The system architecture shows 2.2.3.1 Unit Testing
detailed hierarchies of the entire system, down Unit testing takes place at the part or subsystem
to a component level. By putting these diagrams level the goal of this test phase is to verify that
on the wall in the workshop, the team members each part that a subsystem is composed of meets
could show what components were in produc- its requirements and specifications. Some of the
tion (marked with orange sticky notes), delayed parts of the subsystems were bought but most
(marked with yellow sticky notes), critical (red), of them were made by team members. Verifying
produced (green) and assembled (blue). parts can be done by visual inspection, by meas-
uring dimensions or by stress test to the parts.
This visualization was developed to answer
a need from one of the team members who The responsible of verifying the parts are the
wanted to know more of what was happening engineers that have designed that part.
with the different sub-systems, in order to know
where he could help when he had time available, 2.2.3.2 Assembly Test
e.g. while waiting for carbon-fibre to cure. The assembly test phase was divided into two,
the mechanical assembly test and the engine and
2.2.3 VERIFICATION, VALIDATION the control systemt test. The main goal of this
AND TESTING ACTIVITIES OF THE phase was to test all those Shell Requirements
that are not related to performance and to check
DNVFF2 the interfaces.
One of the main focuses of Systems Engineer- The first testing day was done indoors with a
ing in the SEM project has been to design and smooth surface. The NTNUI let the team use
perform a qualification strategy. The decision of Dragvoll’s sport center. It was an excellent place
making such a strategy was based on the SEM to test the visibility of the car, to check Shell
history where previous teams have had a lack of requirements and also to test the mechanical
testing that made the car less reliable. Besides, subsystems. The team thought that it would be
the DNVFF2 is a complete new car and therefore a good idea to test the car for the first time on a
testing all the components and its performance smooth surface where the car was not going to
is of great importance. be extremely demanded.
During the first day of testing some cones were
The Verification, Validation and testing strategy set in zigzag. At the beginning the cones were
was divided into four different and consequent quite far from each other, the steering was
Figure 2.2.13: Template for the Wall-Architecture. Orange means “in production”, red is “critical”, yellow is “pend-
ing”, green is “produced” and blue is “assembled”
performing well, then the distance between the The complete drive train was tested on test
cones was narrowed and at the 3rd trial one of bench with different loadings at different speeds
the pulley mounting points that are used to sup- to measure efficiency for different operating
port the Kevlar cable of the steering fell off. The points. The test bench was also used to study
surface where the pulley’s mounting point was battery behavior for low battery voltage and over
glued to the carbon fiber was not big enough. current. The DNVFF2 had two different types of
This problem was fixed by increasing the contact batteries, one was provided by Altitec and the
area between the pulley and the monocoque. two others by Gylling.
After the first test, when the car was in the work- The second battery pack from Gylling was or-
shop one of the rims experienced a small crack, dered since the team thought that the one that
the team thinks it was due to some problem dur- was already delivered was not going to be able
ing manufacturing at HPC. See chapter 3.6. to provide enough energy to finish one attempt.
This decision was made based on the tests made
After testing and verifying the car at a smooth at SmartMotors. The issue with this decision is
surface the next step was to use Dragvoll’s park- that it was based on the tests done to Altitec’s
ing lot to test the mechanical properties of the battery, it was seen afterwards that it was not
car under a more demanding surface. The engine necessary to have ordered another battery from
was not on the car yet so the team members Gylling since the first one was able to supply
pushed the car to check rolling, braking and more than enough energy. The cybernetic and
steering capabilities. electric engineers of the DNVFF2 had almost no
time to test and that is the reason why the test
An example of the test procedures that were per- were made for just one of the batteries, the one
formed during the assembly test can be found in that was supposed to be the race battery.
appendix H.
The outcome of these tests was used to discover
The other assembly test part refers to the engine that under voltage protection was needed to
and the control system. The assembly test of prevent complete stop while racing, torque limi-
these two subsystems was held at SmartMotors. tation to prevent under voltage and over current
In order to test the propulsion with the control to prevent still stand while racing. Measurements
system a test bench that is located at SmartMo- were used to find the most energy efficient veloc-
tors was used. Not having an in house test bench ity/torque profile for the given track.
made nimpossible to test as much as the team
wanted as the sponsor was not always available.
SEM 2012 team preparing the visibility check

2.2.3.3 Performance Test parameters and vectors were changed to solve
Tests were performed in Dragvoll’s parking lot, as this issue. The last adjustments had to be done in
it was impossible to get the airport or the army Rotterdam as the starting slope was not known.
installations that last year’s team used, the team
also got the approval to test in Trondheim Havn The maximum speed could not be tested due to
but due to weather conditions and time prob- too little space. Ability to turn at high speed was
lems it was not possible to test the car there. tested for the car, with a satisfying result. Coast-
ing made the turns even smoother as the engine
Propulsion- and control systems were given was not supplying any torque.
most attention during this test, as most of the
mechanical parts had already been tested. The
rear suspension was tested in relation with the
heavy motor as it was not designed to be used
with this one. Starting parameters were tuned to
the drivers weight. The starting parameters were
tested with different starting slopes to check if
the engine was able to supply enough torque.
When the car was driving uphill, the motor had
problems applying enough torque. Some motor
Left: SEM team at Dragvoll’s parking lot
Right: Performance test
2.3
PROMOTION
The whole project was depending on support from
sponsors to survive, and to keep them interested,
different kinds of promotion were done. Events,
newspapers, TV and social media gave both the
team and the sponsors important attention.
DNV FUEL FIGHTER 2
DNV FUEL FIGHTER 2
Graphical Profile
This years graphical profile was a
simplified version of the previous years
logo. The small collibri was still the
main part, but with green colors and a
simplified font.
Promotion Video
Student TV was hired to make a
promotion video for the team and the
project. The result was a 2:29 minutes
video that could be found on the
webpage of the team.
The monocoque was already designed and in production when the design of the interior
started.
Homepage
When deciding what kind of webpage was
wanted, the whole team agree that it was
important that each member could edit it on
their own. In this way one would get more
updates, and therefore more visitors.
Wordpress was chosen as the Content

Management System used on the page. This
would allow all members in the team to have
their own username and password, and blog
posts could be added from anywhere at anytime.
The system would also be easy to tranfer to next
years to, so that they could just continue filling it
with posts.
Events
To create publisity for the for the project, get
new sponsors and to make next years students
interessted in the project, the team was
attending to a lot of events. To stand out on the
events, stand material like table, rollups and
posters were produced. When bringing this along
with the car, the team got a lot of attention.
The main event in the spring semester was

without a doubt the unveiling of the car. This was
done at Solsiden shoppingcenter, and a lot of
press and spectators came around to watch.
53
X.X.X.X TV AND NEWSPAPERS
tekst
TV and Newspapers
To get as much media coverage as possible, the
team hired a person to be responsible for PR
and media. Even though the team got several
TV interviews and a lot of newspaper articles, it
was never close to the goal of 150 articles and 3
appearances on TV.
2.4
ROTTERDAM
Image taken from: http://static.zoom.nl/ 10.06.12

2.4.1 SHELL ECO-MARATHON Figure 2.4.1 shows the track but does not give
any information about elevation. The official track
length is 1630 m but the actual length, obtained
The 2012 competition was held at the sport
by calculations, is 1602 m. A valid attempt will
and concert venue Ahoy, south of Rotterdam.
consist of driving 10 laps around the track in 39
Shell claimed as many as 50,000 people visited
minutes stopping once in each lap; this means
the event during its four day duration. The track
that the average speed has to be 25 km/h. The
surrounds the venue, on parking lots and small
track has some elevations that if not foreseen can
roads. Shell has stated that the competition will
cause invalid attempts or higher energy con-
be held in Rotterdam also in 2013. It is possible
sumption. There is a defined place to stop the car
that Ahoy will be selected as venue again.
but not a well defined stopping line in the track.
Drivers can choose where to stop as long as it is
Every team had their own 5 m x 4 m paddock
within the stopping area defined by Shell. There
inside the main building. This area was also open
was an elavated pedestrian crossing within the
to the public during daytime. The participants
stopping area. If the car stops right before the
could spend the night here, but could not leave
elevation it will not be able to start again, due to
the area and return after 23.00.
high torque. The driver of the DNVFF2 was able
to stop on top of the elevation in all runs.
The paddock quickly became cramped for Urban
Concept cars, as they are bigger in size than pro-
The track was divided in four different sections,
totype cars. Upon arrival, the paddock contained
seen in figure 2.4.1. This was done in order to
a locker, two chairs, a table and a power outlet.
optimize the energy consumption (section 3.8).
Bringing foldable chairs and a refrigerator for
refreshments and food, is well worth it.
One of the main changes that the DNVFF2 has
The 2012 competition started on Monday May
implemented in contrast to the prior DNVFF has
14th. Teams registered and unpacked this day.
been the introduction of a fully damped suspen-
sion. The decision of introducing suspension
2.4.1.1 Track
was made when Shell announced that the race
For the first time in SEM history the competition
was going to be in a street track. The most of the
was held on a street track. All the previous years
other participants were competing with a stiff
that NTNU has participated the race has been on
suspension. However, after examining the track
a race track. All the information concerning the
and realizing that the road is quite bumpy the
road characteristics and quality was unknown for
SEM 2012 team still believes it was a good deci-
the SEM 2012 team. Next year’s competition will
sion to implement a new suspension.
also be held in Rotterdam. Therefore it might be
important to share gained knowledge about the
track.
Figure: 2.4.1: Map of the Ahoy. The track is highlighted in red. The numbers indicate corners and driving direction
DNV FF2 being weighted
2.4.1.2 Technical Inspection Shell Marshalls made sure the Shell logos and
The team leader, both drivers and a fourth team team stickers fulfilled the rules. Also the suitcase
member are allowed to the technical inspection. dimensions were measured.
The SEM 2012 team decided that the cybernet-
ics engineer would be the fourth team member, The next step was to measure the brake effective-
so he could answer the specific battery electric ness. This was done on a 20% slope. Visual in-
questions. spection of the brake system was also performed.
The technical inspection started Tuesday May The next step was the visibility check. It was obvi-
15th. It was important to be one of the first cars ous to the Marshalls that the front visibility was
through technical inspection. This way more time not a problem, so it was not checked. The driver
could be spent fixing the car if it did not pass. The was fully harnessed while the rear visibility was
car entered the queue at 07.00. Seven cars was checked.
already in the queue and more kept on coming.
A Marshall walked behind the car with a stick.
First the car and both drivers were weighed. The driver had to say when he/she first saw the
Then the dimensions of the car were checked. stick and when it was no longer possible to see.
After the visibility check, both drivers had to be to answer the thousands of questions of the Mar-
able to vacate the car in less than 10 seconds. shall. The team was not lucky is this part because
it was the first battery electric car in the inspec-
The following test were related to mechani- tion and it was used by the responsible of battery
cal parts. Marshalls measured the tires dimen- electric cars to brief the other marshall about the
sions. They also made sure there were no sharp information that they need to ask the teams.
edgesinside the car that might hurt the driver.
The lights and the windshield wiper was also The team did not pass the technical inspection
checked. at the first attempt. Questions related to the BMS
of the auxiliary battery was not answered precise
The last step in the technical inspection is the enough. The Marshall was later brought to the
one that studies and verifies the propulsion paddock. After some research the team was able
system. In the SEM team the car is battery electric to answer the question satisfactorily and passed
so the last test was to give information and to the technical inspection.
show the electrical system of the car. This part of
the technical inspection was terribly hard for the
team, luckily the Cybernetic Engineer was there
2.4.1.3 Testing would no longer set reference speed to zero.
On Wednesday, May 16th, the track was open 3.5 Pressing the coasting button triggered a decel-
hours for testing. During the first test the upper eration function. The car was then consuming
part of the door flew open. a small amount of energy. To coast without any
The door was taped as a temporary fix so the energy consumption, the driver had to activate
testing could continue. The car reached its top the brake pedal.
speed of 36 km/h in section 1. High speed during
cornering was not a problem. On May 17th, it was the last testing day and UC
cars had 2,5 hours of testing. A test plan was de-
In section 2 the car shut down. The same thing veloped to test both batteries and the new code
happened twice and also one time in the mid- for the Gylling battery. The plan was to start the
dle of section 1. Due to small elevations on the testing with the Gylling battery to check if the
road the engine needed to supply a higher problem with the BMS was solved. The car run for
torque than expected. After the first shut down, 3 laps with the Gylling battery and the result of
the team was highly concerned that the battery the testing was successful.
might be damaged. The battery was checked and
found to be in perfect condition. Then another Once the team had proved that the Gylling bat-
attempt was done. Again the car shut down tery worked, the Altitec battery was checked. It
in section 2. During the last attempt this day was not able to supply the same energy as the
another battery was used. This ruled out battery Gylling battery. The acceleration was slower and
problems as the car shut down again. the top speed was not reached. After the first
stop, the car was not able to start again. The bat-
Back in the paddock it was discovered that the tery could not supply enough energy to start the
BMS shut off the battery due to high current. car.
There was also a problem with the coasting code,
causing the battery BMS to shut down. It was chosen to use the Gylling battery for the
last test. The car drove12 laps and the battery
It was decided to test both batteries to see which supplied sufficient energy.
one would provide the best chance of having a
successful attempt. The code of the propulsion More information about the batteries is provided
system was changed to reduce the maximum in the K-brief in appendix G.
torque that the engine could supply. Coasting
2.4.1.4 Race result of163 km/kWh. Winds were favourable and
Friday 18th was the first race day for Urban the vehicle did not have to overtake any cars.
Concept cars. 3,5 hours were set aside for racing.
The first attempt was a safe run, where the motor The last attempt ended in failure. One of the
parameters were set to safe values. This gave a steering cords snapped on the fourth lap. The
valid attempt, and more risky parameters could cords were inspected before starting the at-
be used. tempt, and showed no signs of wear. From
inspecting the vehicle and photos taken during
Due to poor preparation routines the first at- the attempts, the team concluded the snapped
tempt was not valid. The door came loose, and cord was a result of multiple factors. Photos show
the driver was forced to end the attempt as she that the outer front wheel does not touch the
could not fix the door while driving. ground in some of the turns. From inspecting the
vehicle, it was discovered that the grease on the
The door was quickly fixed and properly secured, draglink was saturated with dust from the track,
and the second attempt was started. This at- causing the draglink to slide with high friction.
tempt was valid, and the team registered a result These factors – along with the Z-rod (section 3.4)
of 136 km/kWh. in the front suspension – may have contributed
to undesired stresses on the steering cord.
It was decided to to reduce the air gap in the mo-
tor to increase the efficiency. The result from the third attempt became the
final result for the team, giving the DNVFF2 a
Saturday the 19th was the second race day. 5th place in the Urban Concept battery electric
Urban Concept cars had 4,5 hours available for category. 22 teams entered, 20 teams passed the
racing. The first attempt was successful, earning a inspection and 15 achieved a valid result.
3
TECHNICAL
3.1
EXTERIOR
3.1.1 INTRODUCTION
The design of the body and production planning Exterior graphical design was defined and ap-
was completed in autmn 2011. plied by Bilreklame to the surface of the car.
Final models were made and milling of the molds
was executed at Eker Design. Actual layup and To verify the computational fluid dynamics
production was preformed at High Performance results, the exterior has been tested in the wind
Composites in Fredrikstad. A lot of work went tunnel at NTNU.
into the final assambly of the body in the work-
shop at NTNU.
Detailed components, such as hinges, covers,
towing hook and windshield wiper have been
designed and produced. Windows for the mono-
coque had to be produced in-house, and special
equipment was made for this purpose.
3.1.2 STRENGTH ANALYSIS
This work is a continuation of the work per- would give the highest stress concentrations in
formed in the autumn 2011 (Endresen, et al., the car is when the car is cornering and experi-
2011). A more detailed load case and better ence unevenness in the ground. The load values
boundary conditions have been developed. The for this case are given by the calculations from
load cases are reduced to the worst case, and 2011 (Endresen, et al., 2011):
applied in the most critical part of the vehicle. A
layup configuration for the monocouqe has been Driver 70 kg
optimized and tested. The material data used in Cornering 510 N
2011 was changed due to change in production Bump 780 N/960 N
method.
These loads are applied in the left front wheel
3.1.2.1 Boundary conditions well. This is considered to be the weakest part of
To reduce the complexity of the problem, it was the vehicle, due to the door opening and little
decided to reduce the number of load cases and transverse support in the front. The suspension
only focus on the most critical load. The case that loads are located in the wheel and connected to
Figure 3.1.1: Suspension system modeled with RB2 elements, loads and moments are applied in the wheel center
and transferred into the body through the stiff suspension system.
the body through RB2 elements. When applying this load case, the car is fixed in
each hub with free rotating in longitudinal direc-
When simulating the loads on the front left tion, which will represent real testing conditions.
wheel, all other wheel wells are fixed. This is to re-
move the large reaction forces in these areas. The 3.1.2.2 Layup
forces will generate a high moment in the other The layup started with applying two layers of
wheel wells, creating unrealistically high stress DB420 (Appendix B) on the entire body in a
concentrations. The focus will only be on the crisscross pattern. The floor of the car was rein-
front wheel well, therefore, the other wheel wells forced with 10 mm thick Divinycell 80 core to add
are of no interest. To remove any high stress stiffness and to withstand the stresses from the
concentrations in these areas, the wheel wells are driver. Two extra layers of DB420 on both sides
sufficiently fixed. of the core were added to give the sufficient
strength to the underside of the body.
To verify that the car is sufficient durable, the The wheel wells were considered to be the most
load case given by the SEM rules are also tested. critical parts and needed the most reinforce-
ments. A 6 mm Divinycell 80 core was placed on
Driver 70 kg the side walls and total of 4 layers of DB420 were
Roll bar load 700 N laid over the entire wheel wells. The roof was
Towing hook load 2000 N reinforced with 6 mm thick Divinycell 80 core.
Figure 3.1.4 shows the overall layup for the car.
4 mm disp.
Figure 3.1.2: Displacement due to loads during race conditions
8 mm disp.
Figure 3.1.3: Displacement due to loads during testing

6 mm H-80
4 x DB420
6 mm H-80
2 x DB420
10 mm H-80
Figure 3.1.4: Layup for the car, placement of 4 layers (top), core reinforcemenst (middle) and 2 layers (bottom)
3.1.2.3 Results solution can be to make holes in the core mate-

When modeling with carbon fiber, the stresses rial, filling up the holes with several patches of
usually are not the problem. The major challenge carbon fiber.
is to make the structure sufficiently stiff, so de-
lamination, buckling and cracking are neglected. When applying the loads given by the rules, the
displacement in the right a-arm is too high. A
The analysis of the loads from the race conditions solution for improving this can be to further add-
shows a deflection in the whole wheel well and ing more core material, making the arm thicker.
in the hood. Maximum displacement is located An even better solution is to widen the arm,
in the hood, flexing the hood 4 mm upwards. The making it symmetrical to the left side, and have
consequence of this is considered to be small the same laminate configuration. This is just a
since further strength will be added when install- matter of adding more carbon fiber and a wider
ing windows and a dashboard. piece of core material during production. So no
further optimization of this part is considered
As predicted the highest stress concentrations is being necessary. The large deflections around
located around the fixing points for the suspen- the front lights due to the loads from the towing
sion. The analysis shows rather small concentra- hook is not considered to be a problem, since the
tions of stress, but a solution for distributing the area around the placement of the towing hook is
forces to the wheel well wall must be established. sufficient stiff. This case is also an extreme case,
This is to avoid ruining the sandwich structure most likely never to occur.
by drilling holes through the core material. A
Stress concentrations
Figure 3.1.5: Stresses in the most highly stressed carbon fiber ply
3.1.2.4 Conclusion
The FEM analysis of the monocouqe will make a
basis for the production. This work also reveals
critical areas and gives an indication on how the
laminates will behave. The final configuration of
how to implement the different laminates will be
decided in cooperation with the experts in this
field. The final configuration will also be highly
influenced on the expert’s expertise and produc-
tion technique during production.
Figure 3.1.6: A draft angle analysis done in Catia to discover what needed to be improved before the production.
3.1.3 DESIGNING THE MOLDS

When the autumn semester was finished, a from other designers. Since this model should
concept planned for production was made, but it be the base for later making the molds, it was
was still not ready to be produced. important to keep split lines and draft angle in
mind when drawing it.
To make a 3D-model without any errors, it was
decided to build it all the way from the bottom When the process of making the molds started,
again, and then also include the results from the the team got a lot of help from Eker Design. The
last strength and aerodynamic analysis as well as contact in Eker Design, Ketil Humlekjær, gave a
small changes in the shape based on feedback lot of feedback on how the mold would be made,
Lines around the rear hatch
Indents for windows
Lines around lights
Flanges around the edge Lines around the door

in the bottom mold
Figure 3.1.7: 3D model of molds
and what should be changed before production. The final result was a .step-file that was sent to
Some of the draft angles needed to be improved Eker Design, and the production could begin.
(up to 7°), and the split line had to be moved. To
find the split line, a split line analysis was done
in Catia, and then added to the models. (Since
SolidWorks do not have such functions)
To finalise the mold for production, horizontal

flenges were added all around the edges to indi-
cate were to make the cuts. Also small 1 mm thick
lines were added around all cuts, like door, front
and rear lights and rear hatch.
3.1.4 PRODUCTION
In autmn 2011 different production methods were decided to be made with seamless mod-
were considered. The method chosen was highly eling paste. Jackopor would deliver the base
dependent on sponsors and expertise. At the end molds and Svas Kjemi would deliver the epoxy
of the semester in 2011 all sponsors were in place paste. Further information about the production
and the production method of the monocouqe method and molds is given in the project report
and molds was finally set. of 2011 (Endresen, et al., 2011).
For the production of the monocouqe it was
chosen to use vacuum infusion in cooperation 3.1.4.1 Milling
with Paal Fediuk at HPC in Fredrikstad. The molds Before milling could start, the expanded polysty-
rene (EPS) (Jackopor 300), delivered by Jackon, The two blocks were first rough milled down with
had to be glued together to form the correct an offset of 10 mm to form the basis for a hard
size. Since the mold was wider than the standard shell. Secondly, a finer milling was performed to
block size, two and two blocks had to be glued even out the rough grooves from the first milling.
together. The blocks were glued together and
glued to a piece of MDF plate. A plate of MDF After the milling, all the milled off EPS had to be
was used on both molds to be able to fasten removed and the surface was cleaned to ensure a
the molds to the milling machine. Screws were good bonding between the EPS and epoxy paste.
used to attach the plate to the floor in the milling A layer of epoxy paste (P25) was laid over the two
machine. Epoxy glue was used to glue the EPS parts in a thickness of 15 – 25 mm. The epoxy
blocks and the MDF plate. The final result was paste consists of two components mixed togeth-
two EPS blocks of 1555 x 3124 x 920 mm (top) er forming a hard material. This material can be
and 1555 x 3124 x 560 mm (bottom). milled and shaped into the desired form.
Rough milling of EPS blocks at Eker Design in Fredrikstad
Bottom mold after fine milling
After curing of the paste, a new round of rough material. The final part of the top mold had to
milling was performed followed by a round of be laid with P26, a material with higher density
fine milling, resulting in a 10 mm thick layer of and viscosity. There were some concerns that the
epoxy paste. This round of fine milling cut down differences in material properties could create
preparation work before the layup, ensuring a cracks during hardening, but the likelihood of
smoother surface. these concerns was considered too be small.
Since there was not an option in postponing the
Due to some calibration and cleaning of the mold production, the last part of the top was
mixing machine, there was a lot of P25 material therefore laid with P26.
going to waste. This resulted in a shortage of P25
Epoxy paste being applied by Markus Potthoff, a representative from Ebalta
Epoxy paste finished laid at the top part. The difference between P25 and P26 is clearly shown (P25 in light brown)
Milling of epoxy paste
After the milling was finished at Eker Design, of polyurethane paint. This is an important step
the molds were transported to HPC were the to fill all the pores and protect the molds. Some
final preparations before layup was performed. types of paint can react to the epoxy resin mak-
Cavities and imperfections in the epoxy layer was ing the final product impossible to extract from
filled and fixed and a final sanding smoothed out the mold. It is therefore important to use the cor-
the surface. The molds were finalized with a coat rect type of paint for this purpose.
Tracing of lines for the lights was the final step for the milling machine
Finished prepared molds, treated with a layer of blue polyurethane paint at HPC
3.1.4.2 Layup lap of 40 – 60 mm, see figure 3.1.8. By using this

Before the layup could start, the molds were technique, the round edges, which can normally
treated with three layers of wax. The wax was be an exposed area of stresses and strain, will be
applied and then wiped off, three times to ensure stiffer and more solid. These areas would achieve
that every corner of the molds was treated. In twice the amount of layers than the layer beside
some critical areas with complex geometry and the splice. It is important not to position the
steep corners the wax was applied four times. splice on the same place, but always try to stack
The wax will prevent the finished carbon fiber to the overlay like shown in figure 3.1.8.
stick to the mold, therefore it is important to be
accurate when applying the wax.
The first step was to lay a line of vacuum sealer

tape around the edge. It is important that there
are no impurities in the tape edge or on the tape.
The smallest fibers or particles can cause air
leakage when the vacuum is applied. Therefore
the layup process always starts with the vacuum
sealer tape when the molds are clean and free of
any carbon fiber and such.
Figure: 3.1.8: Overlay in a corner, two layer layup
Carbon fiber mats were cut and laid in the molds
leads to four layers in the splice area.
according to the plan made after the strength
analysis. By using textile spray glue it was pos-
For extra reinforcements it was used different
sible to firmly hold the mats in place. It is crucial
types of core material. The core material was cut
not to use too much glue, since this can saturate
out and placed in center between the carbon
the fibers, resulting in weakening the final prod-
fiber layers. To ensure a nice finish on the core
uct. The edges of the carbon fiber were always
material inside the car and to ease the draping,
laid in corners and round edges with an overlay
the edges were rounded with sand paper.
of 20 – 30 mm. Then the next mat was spliced
to the first mat with and overlay of 20 – 30 mm
on the other side of the edge, creating an over-
Waxing of molds
Carbon fiber laid with an overlay of total 40 – 60 mm.

10 mm thick Divinycell core in floor and 3 mm thick Sorix core in the wheel well perimeter
The mounting points for the suspension were and used to mark out the suspension mounting
reinforced with carbon fiber patches, creating a points in core material. Then a round cut of 70
solid carbon fiber piece to mount the suspension mm in diameter was made around the marks,
in. This was done to prevent drilling holes in the creating circular holes in the core material. The
core material, which will weaken and ruin the size and number of layers of carbon fiber patches
sandwich construction. A template was prepared was decided on site.
Side wall reinforced with 6 mm Divinycell thick core and carbon fiber patches for the mounting points
6 mm Divinycell + 2 x DB420
10 mm Divinycell + 2 DB420
2 x DB420
4 x DB420
3 mm Sorix + 4 x DB420
8 x DB420
Figure 3.1.9: Overview of the final layup of the car
The carbon fiber mat used for layup was a ±45°, When all of the carbon fiber and core material
the orientation on the car was not considered to was placed onto the mold, the molds were ready
be that important as long as the mats were laid to be prepared for the resin infusion.
in the most material efficient way. This resulted in
crisscross pattern, ensuring strength in all direc-
tions, and minimum loss in material.
Peel ply applied to the top part, held in place with textile spray glue
First a layer of peel ply was placed over the car- with the same sealer tape.
bon fiber. This is a layer that is easily removed af-
ter curing, preventing resin flow mesh, tubes and After placing tubes for resin inlet and outlet,
plastic bag from sticking to the finished product vacuum was applied on the edges of the mold.
and ensuring a clean surface. Air leakages were discovered with the use of an
air leak detector and sealed with tape. Then the
One layer of resin flow mesh goes on top of the resin was sucked from the center and out to the
peel ply. This is used to easier distribute the resin edges.
through the carbon fiber.
The infusion was stopped by clamping the inlet
At the edges of the mold, the resin flow mesh tube after the resin started to exit all the outlet
was placed 80 mm from the edge. This is to slow tubes. Vacuum on the outlet tubes was still run-
down the resin at the edge, preventing the resin ning after the infusion was completed to extract
to exit before all the fibers are infused. By using air trapped inside the fibers. The parts were set to
multiple exit tubes and clamps, it is possible to cure in room temperature. Plastic bag, resin flow
further control the resin. mesh and peel ply was removed after the curing
process. The bottom and top part was removed
Finally the vacuum bag was placed over the from the mold with the use of wood wedges and
molds. The bag must have sufficient size and brute force. Final stage was to post cure the parts
slack to fit into every corner of the mold when at 80°C for 6 hours. Right after the first curing, the
vacuum is applied. The plastic bag is attached carbon fiber can be soft and pliable. To ease the
to the vacuum sealer tape around the edges. process of releasing the carbon fiber from the
When attaching the plastic to the edge, waves mold, it can be a good idea to perform this task
are added along the edge, to create the sufficient before final curing.
slack that is needed. These are then sealed off
Resin flow mesh, held in place with pieces of vacuum Resin flow mesh at the edges of the mold, the resin
sealer tape slows down considerable
Tubes for vacuum and resin. Resin trap is placed before the vacuum pump, preventing any resin entering the pump
Layup of the back hatch
3.1.4.3 Doors 3.1.4.4 Firewall

The door and the back hatch is both produced The firewall consists of a sandwich structure with
with two layers of carbon fiber without reinforce- two layers of carbon fiber and a 6 mm Divinycell
ments. It was decided in Fredrikstad that the core between the layers. The wall was produced
door should be reinforced, back in Trondheim in a rectangular piece a bit bigger than the
when the door was placed on the assembled car frontal area of the car with the use of vacuum
to ensure a perfect fit. infusion.
Production of the doors follow the exact same

procedure as described for the monocouqe pro-
duction.
Indent made for the window with the use of 3 mm Sorix
Layup of the side door

Water reservoir placed over the mold, creating pressure to the carbon fiber in the mold
3.1.4.5 Wheel well covers The production of the front covers was made in
Since the bottom part of the monocouqe has four parts. The two molds that made up the front
integrated wheel wells it was not possible to use covers on each side were not glued together,
the bottom and top molds to produce the wheel making it easier to fix it on the car on a later
well covers, as done in 2008. Therefore separate stage. Because of the porosity of the mold mate-
molds were produced at NTNU. rial, it was not possible to use vacuum to apply
pressure during production. To apply pressure
The milling machine at IPD is a 3-axis milling to the fibers, a new method was established.
machine capable of milling maximum 70-100 By creating a water reservoir over the mold, the
mm deep. The molds were milled in two separate pressure from the water is used to apply pressure
parts for every wheel well cover. On the sides an to the wet carbon fiber. After filling up the reser-
additional of 40-50 mm was milled to ensure suf- voir with water, it is possible to remove wrinkles
ficient size on the covers. and air inside the fiber with the use of a roller. All
parts of the front wheel well covers were made
Front covers are design to be an integrated with one layer of carbon fiber.
part of the car. To cut costs in production, it was
decided to mill the molds in a cheaper material. For the rear wheel well covers, the two molds
Since these covers only needed to be produced that made up one cover were glued together.
once, and then be fixed on the car, the molds Rear covers are produced with one layer of
could be scrapped after production. carbon fiber twill on the sides (exposed carbon
fiber), one sheet of carbon fiber covering the
The molds were sanded down and filled with whole part and strips of reinforcements on the
filler and then sanded down again. After final edge to further stiffen the cover.
sanding the surface was painted and coated with
release agent. First the rear wheel covers were produced with
wet layup and vacuum, but due to air leakage,
vacuum was not achieved. This resulted in air reduce weight. For the front, the assembly tech-
bubbles trapped inside the fibers and heavy cov- nique will result in a total of two layers, making
ers, these were not approved. New covers were the front part more than capable to withstand
produced with the much simpler and successful the stress of normal use. Lifting or similar actions
water reservoir method. should never occur on the covers or in the wheel
wells. Lifting should always be done onto the
Since none of the wheel well covers will contrib- original monocoque structure on both sides of
ute to the overall stiffness of the car, the parts are the wheel, by four or more persons.
produced with a minimum use of carbon fiber to
Rear wheel well covers
Water reservoir method used to produce rear wheel well covers

3.1.4 ASSEMBLY
3.2.4.1 Monocouqe with Araldite to ensure the position of the firewall
After the doors, firewall, top and bottom part ar- would remain in position during assembly. The
rived in Trondheim, the assembly process began. final assembly of the three parts was done by
First step was to fit the top and bottom part laying strips of carbon fiber on the seams from
together. It was decided to use the flange from the inside. Two layers of carbon fiber were laid
the production process to connect the two parts between the top and bottom part, first one layer
during assembly. The flange around the wheels of 100 mm wide and second a layer of 150 mm
on the top part was cut away and sanded down wide carbon fiber. Between the shell and the fire-
to make a perfect fit to the bottom part. The wall one strip of 100 mm carbon fiber was laid.
firewall was modeled according to the 3D shell Before the carbon fiber was laid, the area around
model and milled at IPM. Final cuts in the firewall the seams was roughen with 60 grit sandpaper,
were done by hand. vacuumed and cleaned with acetone. The fiber
strips was laid with normal wet lay-up, no pres-
First the firewall was glued to the bottom part sure applied.
Gluing of the firewall to the bottom part
Screw connection plates were made to hold the two parts together
Rough cutting of the monocoque. Thin red tape was used to strengthen the marks from the divider, making sure
that the line would not disappear during cutting
After curing and post curing at 80°C for 6 hours, The outer side of the seams and all edges of the
the additional flange was cut away using a pneu- monocouqe and doors were sanded by hand to
matic cutting tool. Since the carbon fiber dust ensure smooth and even edges. The edges and
conducts electricity, the major part of the cutting seams were sanded with the use of rough grit
was done outside with sufficient protective gear sand paper and files. A divider was used to mark
such as respirator, glasses, hearing protection the cutting line, 10 mm from the indent edge.
and two pair of gloves. By using pneumatic tools,
the dust would not be able to ruin the tools, and 3.1.4.2 Wheel well covers
ventilation do not become that critical when The front wheel well covers was cut around the
working outside. Rough cutting the windows and edges and fastened to the monocouqe with
doors was also performed outside. the same technique as for the firewall, with
one carbon fiber strip of 100 mm. The task was
Carbon fiber strips in the seams, holding the three parts together
Gluing of the side parts of the front wheel well covers
performed in two operations. First the side part layers of carbon fiber after fastening the covers to
of the wheel well covers was fastened, the molds the monocouqe. This is more than adequate ac-
was used to hold the side part in the correct cording to strength and stiffness for the covers.
position throughout the fastening and curing
process. The decision to postpone the final cutting of the
covers was made to ensure that they would be
Secondly, the bottom part was fastened, using as tight as possible around the tires. This proved
the same method with support from the bottom to be difficult since the alignment of the front
part molds. suspension is a challenging process and even
harder without being able to mount the wheel. A
In total, the front wheel well covers consist of two template of the wheel was prepared and after a
Both parts of the front wheel covers fastened to the monocouqe
quick alignment of the suspension it was possible covers, the covers were constantly adjusted due
to make marks for the perimeter on the covers. to ongoing tweaking of the rear suspension. The
final cuts for the rear covers were also performed
The final cut of the bottom part was done after in Rotterdam.
the front suspension and steering was in place. Fastening method of the rear wheel covers is
Some modifications to these parts had to be ex- described in section 3.1.7.4.
ecuted during the whole process, mainly due to
constantly fine adjustments on the front suspen- The constantly realignment of the suspension in
sion. The final cut was performed in Rotterdam, front and rear lead to a bigger air gap around the
the day before race day. wheels than what was desirable. But taken in to
account the complex process of fine adjusting
The rear wheel well covers were cut around the a suspension system and eliminating any risk of
edges and fitted to the car. A rectangular slot was scrubbing, the end result is more than satisfying.
cut out for the wheel. As for the front wheel well

Fine cutting of the wheel well covers, Dremel cutting tool and vacuum cleaner is used
Reinforcing the door with 50 mm wide Rohacell 31 core
3.1.4.3 Doors resulted in softening the carbon fiber and the

The edges of the door and back hatch were door losing its original shape. The hinge connect-
roughly cut along with the monocouqe and ing the door to the monocouqe is described in
sanded down by hand. Since the passenger door section 3.1.7.1.
is only made out of two layers of carbon fiber, re-
inforcement was added in the door to stiffen the It was decided not to reinforce the back hatch.
construction. The door was clamped on to the The two layers gave the door some flex, making
monocouqe and the reinforcement was added it easy to fit onto the monocouqe. Carbon fiber
when the door was in place. This would help to bracket were used to mount the back hatch to
preserve a perfect fit to the monocouqe. Howev- the monocouqe. Brackets for the back door is
er, an error was made when the door was placed described in section 3.1.7.3
in the oven to post cure the reinforcement. This

Back hatch fitted to the monocouqe
Rear wheel covers fitted to the monocoque

Finished car, foiling has commenced
3.1.4.4 Surface finish production. After all the gaps, seams and imper-
To create a smooth transition between top and fections were evened out with body filler, all the
bottom part and the front wheel well covers and surfaces were sanded down. The car was sanded
body, the seams were filled with body filler. The down 3 times; first with 120 grit paper, followed
filler was sanded down and new filler was applied by 400 grit paper and finally 600 grit paper.
to fill in the remaining gaps. This process was
repeated until the seams were perfectly smooth. Just before foiling, the surface was cleaned
several times with alcohol to remove any dirt and
The body filler was also applied to areas around dust.
the car was there were imperfections from body

Body filler applied around the front wheel well cover
Sanding of the whole car, a time consuming process

3.1.5 WINDOWS
The aerodynamically design of the body includes object. Because of the door the windshield needs
double curved windows. Indents for windows to be divided into at least two smaller windows.
were included in the negative base molds of the Figure 3.2.10 shows the general shape and ar-
car body. Positive molds for all three windows rangement of these. The front window and side
could then be produced based on the base window could be combined into one, as the car
molds for the body. The SEM rules recommend only has one door. Ideally the shape and size of
windows made by polycarbonate, based on its the front window should have been optimized
impact abilities. It is important that the windows towards production during the design of the
do not shatter into sharp shards if hit by a foreign body. Implementing single curved shape or lift-

ing the frontal part of the window would simplify negative response from most companies. Those
the production process. who initially gave positive feedback were too
busy within the time limits of the project. Deci-
Several companies have been contacted regard- sion was made to do the production in-house,
ing production of windows. The fact that the with the equipment available at IPM and IPD. Size
shape is double curved and the general size and shape of the window required a new vacuum
of the windows introduced some unforeseen table. The solution with three windows was cho-
complications. For production of these windows sen to reduce the size, cost and complexity of the
in polycarbonate, vacuum forming is the only production.
possibility. Molds need to be produced, and the
final result is highly dependent on the surface
finish of these molds. The uncertainties regarding
the optical properties of the final product lead to

Figure 3.1.10: Arrangement of the windows
3.1.5.1 Molds of sanding is usually required. Due to the smooth

The molds from production of the car body are surface of the body mold, the window molds
accurate and precise, it was desirable to reuse were quite good so start with. Fine grade sand-
these in the production of the windows. The paper in the grit range from 120-600 was used to
body mold was applied a release agent at the improve the surface of the molds. Several coat-
window surfaces. Then a thin layer of polyester ings were tested with varying results. Since the
based gel-coat was applied followed by several molds were produced with polyester, the mold
layers of fiberglass in polyester. The polyester was itself cannot withstand high temperatures over
cured overnight and positive molds/plugs of the time. Melting temperatures of polyester can be
windows were retracted from the negative body found down to 200 °C, and the polyester will lose
mold. To secure overlap between the different its structural integrity even before that. Some
molds the production was done in two stages. coatings like heat resistant spray paint needs to
First the molds for the side windows were pro- harden at the working temperature. Local heat-
duced, then the front window. ing of the surface with a heat gun was tested on
these types of coatings. The rapid heating caused
Forming temperatures up to 220°C require strict small blisters and cracking of the surface. Coat-
control over the mold surface. Both texture ings requiring heat hardening above the working
and coating is important properties with direct temperature of polyester was therefore aban-
impact on the final result. The surface roughness doned. For the side windows the gel coat used in
must decrease with increased forming tem- the first step of production was sufficient regard-
perature. The coating must withstand forming ing temperature resistance. Small deformations
temperature while giving a smooth surface. To leading to low forming temperature made this
prevent waste of large plastic sheets, it is recom- possible. On the front window much higher
mended to do small scale testing of the coating temperatures is needed and epoxy was used as
and surface roughness before production. coating. The epoxy left some residue on the poly-
carbonate window which was removed with ROT
To improve the surface finish of the molds a lot WEISS Acrylic- & Plexiglas polish paste.

Gel-coat applied to the front window mold, the layer will provide a high quality finish to the molds
After a quick hardening of the gel-coat, several layers of fiberglas infused with polyester was applied
Vacuum forming of the side window
3.1.5.2 Vacuum Table cordingly high temperatures, longer drying time

A vacuum table made out of plywood was manu- and temperature should be considered.
factured for this project. Two plates of plywood
were prepared in the same size. One plate was Polycarbonate sheets were mounted in a solid
perforated with thousands of 5 mm holes. The aluminum frame that can withstand the tempera-
other plate had air channels milled out to con- tures and forces induced by the forming process.
nect the holes when the plates were joined. It is important to seal of all connections between
Connection for the vacuum was included in the the vacuum source and the plastic. Areas that
same plate. Silicone was added along the edge require special attention is the connections
of the plates, and they were screwed together. between the solid frame, plastic and the vacuum
Along with the vacuum table a frame to support table. During production an expanding sealer
the plastic is needed. IPD had an aluminum frame tape bought at biltema was used.
in storage, and to save money and time this
frame was used. The size of the vacuum table was The frame with plastic was heated up to first
therefore given by this frame. Ideally the vacuum drying and then forming temperature. When the
table should have been even bigger. This would polycarbonate was heated sufficiently it started
lead to more waste of plastic, but could have to sag. The frame was removed from the heat
made the forming process easier. source and pressed down over the mold located
at the vacuum table. Vacuum between the table
3.1.5.3 Production and plastic was created with two vacuum clean-
The door and side window was produced with ers. After several attempts the suitable forming
1mm thick polycarbonate. As the front window temperature was found, and the polycarbonate
is larger and subjected to external load from deformed perfectly around the mold. For the side
the windshield wiper, 1.5 mm thickness was window this temperature was about 180 °C. A
chosen. Over time the polycarbonate absorbs considerable higher temperature was needed for
humidity from its surroundings. If this humidity the front window, due to larger deformations and
is not removed before heat forming, small bub- increased thickness. The most successful attempt
bles will appear all over the formed plastic. This was done at about 220 °C.
reduces the visibility through the polycarbonate
windows, and must be avoided at all times. The Correct placement and fastening of the mold is
producer of polycarbonate sheets used in SEM important to achieve good results. Areas with
2012 recommends drying the sheets at 125°C for low deformation should be placed close to the
about 1 hour (GE, 2012).When forming complex aluminum frame where the plastic is fastened. If
shapes which require high deformation and ac- possible areas with larger deformation should be

located close to the middle of the frame. forming process above the forming temperature
would most likely give great improvements on
Production of windows in polycarbonate with the result. Thickness of all windows can be re-
double curves is a complex process. This is duced, either to 1mm for all windows or even less
important to emphasize as many unsuccessful at- for the side and door window.
tempts were made. Sufficient time and resources
must be planned for this part of the project. Since 3.1.5.4 Assembly
there is little experience and expertise to be The windows were cut out from the plastic
found at the university, a professional company sheets, and fine adjusted to fit the body. First
should be contacted early in the autumn semes- the front window was glued in place. Araldite
ter. To further simplify the production process 2031 was applied on the indented connection
other plastics should be considered. Polycarbon- lip of the body and pliers were used to hold the
ate scratches quite easily, and mounting, clean- window in place. The glue was cured overnight
ing and operation of the windshield wiper will and the pliers removed. Next glue was applied on
over time leave marks on the windshield. Switch- the body and the side window was put in place.
ing to acrylic plastics could solve some of the Tape and wooden sticks was used to keep the
scratch problems. In addition forming without windows in place, while applying pressure on the
the need of vacuum might be a possibility. Molds connection area. When the araldite had cured,
made from a more suitable material should be Tec-7 was applied to create a smooth transition
considered. The ability to perform the entire between the body and the windows.
Milling of air channels in the vacuum table, connecting Windows assembled on the car
the holes drilled out in the ply wood
The lights of the former car was vacuum formed in plastic, and used three different LED-lights for high beam, low
beam and indicator lights.
3.1.6 LIGHTS
As a normal car, the Urban Concept cars in Shell tic had melted, but everything did still work as
Eco-Marathon needs to have the same lights that it should. The front lights were attached to the
are required on the streets body with velcro on the inside.
3.1.6.1 Former Lights

The former lights were fulfilling the requirements
by having a separate star-LED with lenses on
each of the blinkers, high beams and low beams.
In the back the car had a strip of red LEDs as
break indicators. The front light glass’ were made
of PET, as well as the inside of the lights. Because
of the heat from the star-LED some of the plas-

Figure 3.1.11: The holes for the lights needed to be planned before the production of the monocoque started, and
therefore also before the inside of the lights were designed.
3.1.6.2 Interfaces 3.1.6.3 List of Requirements

The only interfaces one would need to consider Both the rules and the team goals set require-
when designing the lights, were the monocoque ments for the lights:
and the control system. Holes for both front
lights and rear lights needed to be planned be- Rules
fore the production of the body started, to make • Must have two front headlights
the process of cutting the holes a lot easier. • Must have two front turn indicators
• Must have two rear turn indicators
The control system were depending on voltage • Must have red break lights in the rear
of the lights, and therefore also the number of di- • Must have two red rear lights
odes. Because of these interfaces the light design • The center of each headlight unit must be
needed to be finished early in the process. located at an equal distance and at least 30
cm from the lonitudinal axis of the vehicle.
Team goals
- Each light must not weigh more than 150 g.

3.1.6.4 Materials and Technology The diods selected were of two different types;
There were two obvious choises of material for Star LEDs for the high-beams and the blinker in
the lights; either carbon fiber or vacuum formed the rear, and 5 mm LEDs in different colors for
plastic. Each of the choises would have advan- low beam, blinkers in the front and driving lights
tages. in the back. The 5mm LEDs would not emitt any
heat, while the Star LEDs should be isolated from
The carbon fiber would resist the heat from the the plastic.
LEDs much better, but would be the hardest
one to produce. The plastic would on the other
hand probably be both the lightest solution and
the one that was easiest to produce. The heat
problem could be solved by insulating the LEDs
better.

Figure 3.1.12: Different combinations of light placement were tested in SolidWorks.
3.1.6.5 Concept Development The sketches and ideas could later be combined
Since the overall shape of the lights were into CAD models in solid works, where it would
developed last semester, the work that had to be be easier to plan and place LEDs. In the end all
done was to figure out how to finalize the design the best ideas were combined into one finished
and make it possible to produce. concept.
The initial sketching were done on printed CAD

drawings of the front and rear of the car. In this
way it was possible to generate a large amount
of ideas in a short time, which was crucial since
the design needed to be finished in a very short
period of time.

Rear low beam
High beam
Indicators
Indicators
Low beam
Figure 3.1.13: The final front light and rear light concepts.
3.1.6.6 Final Concept used. The indicator lights are orange Star LEDs
The final concept basically looks like last semes- palced between the strips.
ters concepts, but have diods as light sources
instead of the laser lights. The front lights have All the LEDs are mounted on a vacuum formed
strips of LEDs in top and in the bottom, and in plastic part, made of spraypainted PET. The whole
between one can find the indicator lights, which light is mounted on to the car with velcro.
also are 5 mm LEDs. For high beams, one Star LED
is place on each side of the car, in the middle of
the two low beam strips.
The rear lights are quit similiar to the head lights,

it the way that they also got the strips of LEDs
along the top and bottom, just that red lights are

8 differnent molds were made to vacuum form all the The inside of the lights were vacuum formed in transparent
parts for the lights PET, while the outside were made of poly carbonate.
The final front and rear lights mount in the finished car. The lights are attached to the car with velcro on the inside.
3.1.6.7 Production To finalise the lights, the diodes were glued in

Molds for both the glass and the inside of the and soldered together.
lights were designed, and then milled on the CNC
mill at IPD. The molds were then sanded down
and painted, to get a perfect surface, since every
single scratch would leave a mark in the glass.
When the molds were finished, the glass were

vacuum formed with 1 mm poly carbonate, and
the insides were formed in 1,5 mm PET. Holes for
diods were drilled and the whole thing painted in
black.

3.1.7 DETAILS
The design of DNVFF2 and SEM rules requires proper attention. This has resulted in fast and
many small detail components. These have to be simple solutions that in most cases lack the feel-
designed and produced along with other major ing of robustness. The sturdiness of the door has
parts. Saving weight in these small and relatively not been good enough. Specifications from the
simple but numerous components will help to autumn semester (Endresen, et al., 2011) require
reduce the total weight of the car. SEM rules proper design of all detail components. Robust-
states that the urban concept cars should be able ness and a general feeling of quality has been
to drive in light rain. Sufficient ventilation and an the goal for the components mentioned in this
effective windshield wiper are mandatory. Earlier chapter.
years the detail components have not been given

3.1.7.1 Door Details one have to open the door to reach the specified
Both door hinges and door handle are depend- door opening by the SEM rules of 500x800mm.
ent on the placement of the door and in what Simple CAD models of selected concepts are
direction it opens. Straight connection lines on shown in figure xx. These were made to further
planar surfaces should have been implemented investigate the different qualitys of the solutions.
in the design of the body. This would simplify
both the design and production of the door Gullwing doors shown in figure 3.1.15 are
details. The design of DNVFF2 does not include mounted in the roof of the car. This was initially
such design properties. Several different place- the planed solution when designing the body.
ments of the hinges shown in figure 3.1.10 were It looks good and gives the driver good access
therefore thought of. to the car. Reaching the required door opening
is not a problem with this concept. Since the
Another important factor is how many degrees window is not stiff enough to carry any load, the

Figure 3.1.14: Different placements of the hinges
Gulwing
Lambo
Suicide
Figure 3.1.15: Different door opening solutions

frame around the window must be reinforced. opening according to rules. This solution has the
A separate mold would have to be produced to advantage of the door staying open in most posi-
manufacture the door. Connection line between tions by itself. The concept of suicide door was
the body and the door are quite curved in the chosen as it would not require a separate mecha-
roof, this increase the complexity of the solution. nism to stay open or reinforcements of the door.
The door will not stay open by itself and a sepa-
rate mechanism is needed for this purpose. The Hinge
gullwing door was discarded for these reasons, Earlier years NTNU SEM teams have used stand-
but should be looked further into. ard hinges from kitchen cabinets made in steel.
The new requirement regarding stability and
Lambo doors shown in figure 3.1.15 are mounted robustness, gave the need for a custom hinge.
in the front part of the door. This concept takes Designed to be as light and stabile as possible,
advantage of the relatively straight connection many concepts were considered. In figure 3.1.16
line between door and body at this position. To some of the iterations are seen.
reach the required door opening the door must
be opened approximately 120 degrees. The front A suicide door solution required the hinge to be
of the door would come in conflict with the opened 180 degrees. To reach the specification
windshield, and this solution was therefore dis- from 2011 (Endresen et al., 2011) and still main-
carded. It is not known whether the door would tain low weight, two small hinges was used. Each
stay open by itself, or needs a separate mecha- hinge weighs about 50 grams, and is produced
nism like the gullwing door. in carbon fiber and aluminum. The final design
takes advantage of two pivot points to secure
Suicide doors shown in figure 3.1.15 are mounted 180 degree movement as shown in figure 3.1.18.
in the back part of the door. This solution takes
advantage of the straight connection line be-
tween body and door. Door movement up to
180 degrees is needed to secure correct door
Figure 3.1.16: Idea generation of different hinge concepts

Final hinge design assembled. Two pivot points ensures 180 degrees opening. A combination of carbon fiber
and aluminum results in a robust and light weight product
The arms of the hinge were milled out from a 2

mm thick plate made with prepreg carbon fiber.
Araldite 3021 was used to glue the carbon fiber
arms to the aluminum tube. To reduce wear it
was important that all moving connections had
metal to metal contact. The aluminum tubes are
designed to completely separate the carbon fiber
from moving components. This will ensure maxi-
mum stability over time.
Araldite was used to glue the hinge brackets on

to the door and body. These surfaces are curved,
and measures were taken to ensure correct
alignment of the hinges. While gluing the upper
and lower brackets were connected with rods.
Wedges of araldite had been prepared to per-
fectly space out the brackets. Figure xx shows all
the components in the hinge assembly including
the araldite wedges. Figure 3.1.18: Principle drawing of the hinge show-
ing the opening motion

Figure 3.1.17: Idea generation of different door handle concepts
Spring loaded door handle made in aluminum safetly secures the door in place. A flush finish ensures little
interference with air flow.
Door Handle with Araldite.

Along with robust details, it is important to keep
the outer surface of the car as smooth as pos- Final design and its function are shown in figure
sible. The door handle must therefore be located 3.1.19. The front part of the handle is pushed in
inside the car, or kept flush with the body. In the by the user, this reveals the back part now com-
design phase low weight, robustness and smooth ing out from the car body. Further movement of
transitions were the most important factors. To the handle will engage the door latch releasing
secure the door in closed position, a latch operat- the door from the body. All components except
ed by the handle was introduced. Some varieties standard fasteners and springs were machined in
of the door handle and latch examined during aluminum. The brackets were glued to the door
the project are seen in figure 3.1.17. with Araldite.
Final design and its function are shown in figure

xx. The front part of the handle is pushed in by
the user, this reveals the back part now coming
out from the car body. Further movement of
the handle will engage the door latch releasing
the door from the body. All components except
standard fasteners and springs were machined in
aluminum. The brackets were glued to the door Figure 3.1.19: Door handle in the open postion

Figure 3.1.20: Idea generation of windshield wiper
3.1.7.2 Wiper Design

For the first time in SEM history, all cars in the ur- The idea of having only one component outside
ban concept class needs an effective windshield the car was introduced quite early in the design
wiper. The type of wiper movement and place- process of the wiper. Combined wiper arm and
ment had to be established. If the windshield is blade could reduce weight and impact on the
big or has a square or rectangular shape, the par- airflow of the wiper. Because the windshield is
allel wiper movement seen in figure xx might be double curved this type of wiper would have to
the best choice. As DNVFF2’s windshield is curved be quite flexible. The stiffness needs to be perfect
in all directions and far from rectangular in shape, along the length of the wiper to secure contact
the normal pendulum motion shown in figure xx between the blade and window at all times. A
is more suitable. The wiper motor and shaft can prototype of this concept was produced, but the
be placed all along the outer edge of the wind- performance of the wiper was not convincing.
shield, but needs some space on the inside of the The uncertainties around the practical testing of
body. The wiper was therefore mounted in center SEM rule regarding the wiper in technical inspec-
of the lower part of the windshield. The wiper tion made the team aim for a safer more stand-
arm and blade should be positioned vertically ardized solution.
in the center while driving. Aerodynamically this
will give the least interference with airflow over Figure 3.1.20 shows some early concepts devel-
the body. oped for the wiper motor. Along with weight the
simplicity regarding programming and control
system was valued. The concepts called wheel
and strut and wheel both have the advantage of
mechanically providing the pendulum motion.
To reduce weight it was important to examine
the possibility to produce small parts. Gears to
slow down and increase the torque of the elec-
trical engine were required. With no possibility
to effectively produce the components needed,
buyable solutions was looked into. Conventional
windshield wiper motors are usually over dimen-
sioned and too heavy for this use. The focus was
therefore turned towards headlight wipers. These
Figure 3.1.21: Different wiper motions motors turned out to be in the correct weight
range, and a used headlight wiper motor from a
Volvo 340 was bought.

Final windshield wiper design with wiper motor, wiper blade and wiper arm.
Production ed at the middle of the wiper blade was quickly
Most parts produced for the windshield wiper are designed.
machined in aluminum. The mounting bracket is
made from prepreg carbon fiber with the same Strips of carbon fiber from the prototype were
mold used to produce the horn bracket. This type heated up to 110 °C. Each strip was pressed down
of standardizing components saves a lot of time on a mold giving them the desired s-shape in
in both production and planning. By stripping one end. The steel plate used in the prototype
down and replacing steel parts with lighter com- production was rolled once more to increase the
ponents, the weight of the motor was reduced by curvature. Four strips of carbon fiber were glued
40 percent. together with Araldite, fixed to this new mold
and reheated. After hardening of the araldite,
To produce the first prototype of the wiper arm a the semi-finished wiper arm was grinded down
mold was made from a steel plate. The plate was to its final shape. It was important in this step
rolled to the desired curvature and supported to constantly check the stiffness of the arm, and
by a plywood frame. By producing the arm in not remove too much material. As a final touch
carbon fiber, the idea was to alter the stiffness the wiper arm was sanded down with fine grade
throughout the length of the arm by the amount sandpaper, and clear coat was applied.
of layers. A 50mm wide sheet of carbon fiber with
thickness ranging from 1-2 mm was produced Assembly
with the water reservoir method. 10 mm wide When all the parts had been produced the
strips were cut out from the sheet, were one of bracket was glued to the car body with Araldite.
them had a 20mm in diameter mounting surface After the glue had cured, the wiper motor was
in the thickest end. The stiffness was tested on assembled and connected to the bracket. The
the mold for the front window. Several strips altered wiper blade was fastened to the wiper
of different lengths were connected with tape arm, which was connected to the shaft of the
to find the optimal thickness. As the window is wiper motor with two M3 set screws. It is worth
double curved this appeared to be more difficult mentioning that during the technical inspection
than first thought. The concept of combined of SEM 2012, the inspectors commented that this
wiper arm and blade was abandoned. A more wiper was the most effective they had seen in the
conventional solution with a wiper arm connect- competition.

Figure 3.1.22: Idea generation for mounting the back hatch
3.1.7.3 Back Hatch Mounting The brackets performed very well in mounting
Attachment of the back hatch was considered to the back hatch. The flexing of the door and the
be easy and development of a mounting fixture tight fit of the brackets made a perfect attach-
was postponed until final assembly. Many dif- ment. During racing, strips of foil were added
ferent concepts were considered, but the desire around the seam of the door. This was to make
to keep the surface clean and smooth gave a lot the door waterproof and to improve aerodynam-
of restrictions. In figure 3.1.22 some of the early ics.
concepts are seen. A new SEM rule also stated
that all covers and doors must be mechanically
attached, so the old solution with Velcro and a
tight fit was not adequate for the new car.
To maintain the smooth transitions and clean

look of the exterior, an internal mounting was
the best option. The bracket solution could use
the flexing properties of the back hatch to keep
it in place. It was decided to produce the brack-
ets in carbon fiber. Two steel plates were milled
down to form the molds. Four layers of carbon
fiber were placed between the molds, clamped
together and the carbon fiber was set to cure.
In addition to the bracket, a screw connection

was placed on top of the cover. This was to pre-
vent the cover form falling off during racing. A
small aluminum disk was threaded and glued on
the monocoque indent to make the back part of
the screw connection.

Figure 3.1.23: Idea generation for mounting the side covers
3.1.7.4 Side Cover Mounting mate foil were applied around the seam before
Attachments of the side covers were also post- racing. This further improved the transition and
pone until final assembly. A solution to attach attachment between the monocoque and side
the covers was not established until a couple covers.
of weeks before departure, due to other more
prioritized tasks. As for the back hatch, a smooth The solution proved to be sufficient and no prob-
and clean look was the most important factors lems were experienced during testing and racing.
when designing the mounting fixtures. Different Assembly and disassembly of the side covers can
concepts were evaluated and a final concept was be a bit difficult. The screw connection is not easy
chosen. See figure 3.1.23 for the different con- to access, but sliding under the car when placed
cepts evaluated. on the ground makes it easier. A new solution to
improve assembly can be developed, but since
A solution with the use of rubber bands and the covers are just essential during race , this
screws was the solution chosen. Two screw con- should not have a high priority.
nections were placed on the underside of the
covers. Threaded aluminum disks were glued
onto carbon fiber brackets and glued onto the
side of the wheel well.
The side part of the side covers were attached

with rubber bands. Aluminum fixings were pro-
duced and glued onto the side covers and wheel
well. Carbon fiber brackets were also produced to
stop the cover from sliding into the wheel well.
In addition to the two screws and three rubber

bands per side cover, strips of green and black

Top: Rubber band connceting the side cover and wheel well.
Middle: Aluminum fixings on side cover and wheel well. Carbon fiber stopper used to secure the side cover.
Bottom: Screw conncetions securing the cover in the tunnel, underneath the car.
Figure 3.1.24: Idea generation for the towing hook
3.1.7.5 Towing Hook The towing hook was milled down to the correct
A new towing hook was design and produced for dimensions and a rectangular slot was prepared
the new car. The old towing hook for the DNVFF in the monocoque. Araldite glue was used to
consisted of a steel threaded bar formed in a U- fasten the hook to the monocoque.
shape, attached to a steel plate glued inside the
car. Reusing this solution was never considered. The towing hook has never been tested in real
There was improvement potential concering conditions or but should withstand loads up to
weight, looks, strength and aerodynamics. 2000 N. Final towing hook weighed 40 g, com-
pared to the old towing hook of 120 g, is this a
To keep the weight down, the towing hook great improvement. The hook is placed in the
should be produced in aluminum. Because of the split line, standing horizontally outwards. This
material properties of aluminum, welding was is the placement where the hook has minimum
not an option. With a T-profile in aluminum the influence on the airflow. The car should never
towing hook could be produced in one piece. be lifted in the towing hook, since the hook is
designed to handle horizontal loads only.
Figure 3.1.25: Strength analysis of the towing hook. The 2000 N load yield maximum stress of 110 MPa.

Top: Towing hook of the DNVFF, threaded bar solution.
Bottom: Final design of the towing hook installed to the DNVFF2. One aluminum part.
The former car had been painted several times, which have resulted in a rough surface
3.1.8 FOIL
The surface of the car is important for the perfor- the years, four layers of paint have therefore been
mance of the car. Just surface friction could have put on the car, which in the end have added a lot
a large impact on the result, so the team was of weight. This problem would be avoided if foil
determined to put a lot of work into making it as had been used all the way, as well as the surface
perfect as possible. would have been smoother. Based on these argu-
ments, the team desided to foil the new car.
3.1.8.1 Paint VS Foil
The former car have been painted every year,
exept the first year when it was foiled. Through

3.1.8.2 Inspiration challenging and interesting foil, but in the end it
The goal from the beginning of the design was the eco cars and electrical vehicle that was
process had been to design and make a car that the most inspiering. Since the car actually was a
looked like a normal street car. Since our car was racing car, an idea formed to combine features
based on a monocoque structure, it did not have from the racing cars with what one often can see
any split lines as the cars we are used to. To give on the electrical cars. Together these expressions
the cars lines, and to make it more dynamic, we could create exactly what was wanted.
were depending on the features of the foil.
A large picture collection of cars that we liked,

and that had interesting foiling features, helped
us starting to get ideas on how to foil the car.
Sport cars and concept cars often have the most

It was important to identify which areas that needed to be foiled and which areas could be left with the original
carbon fiber finish.
Idea generation
The ideation process of the foil had to be done really
fast, the desition to print drawings of the car and
sketch on these would be the fastest way to do the
job. When a large number of drawings had been
produced, selected ideas were tested in solidworks
to see the result properly.

Final concept
The final concept was one of the simplest ideas that
was generated. The main color of the car ended up
as matt black, while detailes were done in bright
green and glossy black. The Fuel Fighter logo should
be cut out of the foil on the hood and the rear to
show some of the carbon fiber structure. A glossy
black bird should be placed on each side.
A normal car usually have four wheels, and so

do this one. But it does not look like it, since the
rear wheels are covered. To solve this problem we
wanted to indicate the wheels with the foil. Green
foil were therefore added around the wheels to give
it the feel of a street car.
The foiling job was done by Trondheim Bilreklame
3.1.8.2 Production and Mounting surface that was wanted. All edges and scratches
The printing of the foil was done and sponsored were patched with foil, so that the final result was
by Printing in Oslo, while the mounting was an almost perfect surface with minimal aerody-
sponsored by Trondheim Bilreklame. Illustrator namic friction.
files of the material was sent to Printing, and
when the car was transported to Bilreklames
graage. The foiling job was done in less than a
day.
The foil is sticky and elastic, and to mount it on

the car, it first needed to be cut into strategic
parts. Later it could be “stretched” around the
car, and somewhere heated to get the smooth

3.1.9 WIND TUNNEL TESTING
Wind tunnel testing was performed on the land and Lien(2010) as a basis.
DNVFF2 in spring 2012. This was carried out by a
group of students lead by one of the team mem- 3.1.9.1 Wind Tunnel
bers. Because of the size and work of this subject, For wind tunnel testing in a closed tunnel, block-
the results are mainly used to verify the CFD from age is one of the main error sources (Eikeland
2011 and a comparison of the old car. Some small and Lien, 2010). The car itself will block a certain
changes to the exterior were also executed to see part of the test tunnel, creating an acceleration of
for some improvement areas. the air around the car. This acceleration will lead
This chapter will use the work carried out by Eike- to an overestimated drag. A correction formula

can be used to correct the error. By using equa- obtains a formula for correcting the blockage of
tion of continuity, a correction factor can be the wind tunnel.
obtained.
In real conditions the car is moving relatively to
the ground and the air is standing still. In the
Where V (true) is the actual speed around the car, wind tunnel at NTNU, the floor is standing still,
V (indicated) is the air flow speed, S is the section creating an error due to the boundary layer.
area and A is the projected frontal area. There are several methods used to correct the
boundary layer; moving belt, ground board and
By solving for V (true) and inserting the drag suction. The cheapest and easiest method is to
coefficient use a ground board, where the car is lifted out

of the boundary layer. Moving the ground with based program called Genlog. After the base
the use of a belt is the most exact method, but reading is completed each individual load cell
can be a very complex process. Applying suction has to be calibrated. Three different weights (0.5
under the car to reduce boundary layer is easier kg, 1 kg and 2 kg) are in turn placed on top of the
and used in many wind tunnels. load cell and a reading is made using Genlog. Af-
ter all 27 readings have been made (the readings
3.1.9.2 Experimental Setup increase linearly) the calibration factors for the six
Testing was carried out in the wind tunnel at load cells can be read from a excel sheet. Lastly
NTNU. The car was placed on top of two metal the air speed sensor has to be calibrated. This is
supports placed on a scale. The scale consists of done by running the wind tunnel fan at certain
six electronic load cells, three for vertical loads revolutions per minute (rpm) and sampling the
and three for horizontal loads. data using Genlog.
Before any testing can commence the equipment When calibrating the wind tunnel for the DNVFF2
in the wind tunnel needs to be calibrated. There there were some problems. The sensor signals
are two main measuring systems that needs to displayed random interference on all channels,
be calibrated; the load cells for measuring drag, and a lot of time was spent on trying to identify
lift and moments, and the pitot/monotube sen- and eliminate this. Different amplifier channels
sor for measuring the air speed in the tunnel. were tested, signal filters were used and even a
All sensors are routed through amplifiers and power supply filter was employed, all to no avail.
optional filters to get more accurate and visible The team even searched the building for any
readings. These amplifiers have to be tuned to equipment that might have been causing the
get the signal within a desirable range, so that interference, but could not find anything. In the
it always can be visually checked during the end the only solution was to run very long sam-
calibration process. A base reading has to be per- ple periods, which meant the calibration process
formed at the start of the calibration process, this took a very long time to complete. To shorten
is done with the test specimen securely in place the calibration time only the horizontal load cells
and the wind tunnel fan turned off. The program measuring drag were calibrated and the rest of
used to read and sample values is a LabView the calibration factors were taken from a previous

experiment. The two calibration factors acquired winds. In the end only three cases were tested:
corresponded very well to those of the previous Base, blocked tunnel and diffusor guide vanes.
experiment and after consulting with students at The car was also covered in small tufts made of
department of Energy and Process Engineering wool to get a visualization of the flow around the
the calibration factors were deemed OK. car.
For the actual test process another LabView 3.1.9.3 Results

based program called ForceLog was used. This Input data
program displays drag, lift and various moments
in Newton directly, and also calculates various Air dentisty 1.23 kg/m3
coefficients if the frontal area of the specimen
Projected frontal area 0.90 m2
has been specified. It also controls the load cell
platform which can be rotated to test for side Test section area 4.86 m2
winds. Due to lack of time no side wind scenarios
Continuity correction 0.664
were tested.
Table 3.1.1: Input data for calculations of lift and
The scope of the work was to test as many drag
aspects about the car as possible, but most test
scenarios where scrapped due to lack of time. In
addition to testing the aerodynamic properties All test scenarios was carried out with the same
of the base setup of the car, it was planned to air speed of 33.7 km/h. The results from the wind
test the effects of several different aerodynamic tunnel testing are displayed in the table 3.2.2
tweaks. It was also desirable to see how or if (corrections made for drag and lift coefficients).
the weight distribution of the car altered with
the speed and how the body performed in side
Test scenario Lift [N] Drag [N] Cl [-] Cd [-]
Base 3.88 9.03 0.08178 0.19029
Blocked tunnel 3.51 11.81 0.07397 0.24887
Diffusor guide vanes 2.18 8.59 0.04594 0.18102

Table 3.1.2: Results from wind tunnel testing

The car during wind tunnel testing. Tufts located over the top remained still, tufts flapping at the sides due to flow
separation after the open front wheels.
3.1.9.4 Conclusion top of the body, and the only signs of turbulent
Wind tunnel testing verified the results from the flow were behind the front wheel wells (as ex-
CFD analysis performed in autumn 2012. The CFD pected). This flow separation can be reduced by
analysis gave a drag force of 6.33 N when travel- developing efficient front wheel covers.
ling at 30 km/h. Compared to the results from the
wind tunnel testing, which gave a drag of 9.03 N Covering the tunnel geometry proved to have
with air speed of 33.7 km/h. Measured in same a negative effect. The drag increased and lift
speed, the result from the testing would give a decreased. But due to the wind tunnel testing
drag of 8 N. configuration, the decrease in lift can be proven
An increase of 20 % in drag from the CFD to the to increase down force in real conditions.
wind tunnel results is as expected. The details on
the real car are much more complex than for the Diffusor guide vanes proved to have a positive
simplified 3D model. All seams, unevenness and contribution to the drag. But more thorough
misalignments of doors and covers will all lead to analysis of the effect and placement must be
an increase in drag. These are all factors that were executed before these changes can be imple-
not considered in the CFD analysis. mented.
The large variance in lift is mainly due to lifting
the car higher in the wind tunnel to eliminate It is also important to note that the aerodynamics
the moving road problem. An increase in ground counts for only about 50 – 60 % of the total resist-
clearance will increase the lift. The lift is also in- ance in the car when travelling at 30 km/h. With
fluenced by the fact that the ground is stationary, the recent changes in the competition over the
moving more air over the car, creating lift. years with more stops and turns, 30 km/h is the
maximum speed used only for small parts of the
The results from the wind tunnel testing show track. Therefore, large aerodynamically changes
that the car has a very low drag coefficient. There which will increase total weight can be proven to
was observed very little flow separation over the be negative for the fuel efficiency.
car. The flow was completely attached over the

3.2
INTERIOR
3.2.1 INTRODUCTION
All the way from the start of the project, perfor- done over a short period of time with few design
mance had been the main focus. Aerodynamics iterations. To develop parts that could affect the
and weight had allways been prioriticed before performance of the driver, like steering wheel
looks and userfriendliness, and all resources had and seat, was then put in focus. Ergonomics and
been put into finishing the body in time. The de- weight was now the most important factors,
signing of the interior was therefore postponed while styling was not prioriticed.
until the beginnining march, and had to be
The monocoque was already designed and in production when the design of the interior
started
The dashboard in the DNVFF was made of pre-preg carbon fiber, and was attached to the car with two screws to
each wheel well
3.2.2 DASHBOARD
To have a dashboard is not a requirement from car stiffer. The top surface was effectively hiding
the Eco-marathon rules, but it would help cov- most of the components in the front of the car,
ering all the mechanical and electrical parts in and making the overall impression a lot tidier.
the front of the car, as well as being a mounting Most of the buttons used by the driver during the
surface for buttons that did not fit on the steer- race were mounted on the dashboard instead of
ing wheel. the steering wheel. This would make the driver
change the grip everytime a new command
3.2.2.1 Last Cars Dashboard needed to be initiated. Sometimes this would
The dashboard in the former car was made of probably affect the drivers performance.
pre -impregnated carbon fiber with a foam core.
It was attached to the wheel wells on each side The dashboard was attached to the wheel wells
of the cockpit, and did in addition to being a with screws, and could therefore be taken out
dashboard, work as a cross beam to make the anytime if reparations or adjustments needed to
be done. This could be useful both when working Control System
on the dashboard, and when when working on All cable from the steering wheel and from the
the mechanics and electronics in the front. buttons on the dashboard needs to be organised
and led to the control system.
3.2.2.2 Interfaces
When designing the dashboard, the interfaces to Monocoque
four different sub-systems needed to be conci- The dashboard needs to be attached to the
dered. monocoque in some way. Preferably to a rigid
part, like the wheel wells to support the steering
Steering Wheel wheel in a best possible way, but it could also be
The dashboard needs to be able to support the attached to the floor or the shell.
steering wheel and keep it stable all the time
while driving. In addition to the steering rod, Steering
the cables from the steering wheel needs to go Must be a bridge between the steering wheel
through the dashboard, either through the rod or and the steering system.
in another way.

Figure 3.2.1: Before the design of the interior could start, one needed to establish a layout of the cockpit, and in that
way set an outline for the size of each component
3.2.2.3 Ergonomics weather. When it is sunny it is important that all

In a car like this, the perfomance of the driver displays are either angled in a way that do not
could count up to 50% of the result, so the make uncomfortable reflections, or that they are
ergonomics in the cockpit environment is really shielded from the sun.
important.
The dashboard should support a correct height
Even though the driver should always focus on and distance for the steering wheel, and at the
the road and never look down, she should be same time not block the view for the driver. All of
able to see all necessary buttons at any time. this should be customized for the selected driver,
They should never be covered by the steering but possible to use for most people.
wheel, mirrors, hands or other disturbing
elements. At the same time, buttons that are Buttons that are used before and after the
used when racing, should be within reach race, and during testing should be placed on
without moving out of a comfortable driving the dashboard. The rest should be placed on
position. the steering wheel to optimize the drivers
performance.
The car should be ready to drive in all kinds of

To give the driver proper feedback when pressing the different buttons, it was important to be selective when
choosing buttons. Important factors were lights and tactile feedback
3.2.2.4 Analysis of Functions • LED-control.

A lot of factors have an impact on the placement • Blinker indicators.
of the buttons. The diameter of the steering
rod limited the amount of wires that could go Used during race
throught it. And of course the space on the • 6 engine control buttons.
steering wheel itself would be a limiting factor. • Horn.
• Low-beam.
The most logic way to devide the functions • High-beam.
would be to place the least used buttons on the • Indicator (left/right).
dashboard, and the most used on the steering
wheel. The result was two different categories:
Used before race

• Emergency shut-down.
• Start/stop.
• Wind shield wiper.
• LED-display.

3.2.2.5 List of Requirements should be reachable from the outside.
The list of requirements can be devided into four • Must include two indicator lights (left/right)
different categories based on whether they are • Must include one emergency shut off button
demanded by rules, users, team goals or other • Must include one LED-display with a separate
sub-systems. control button.
• Must include one wind shield wiper button.
Rules
• The vehicle interior must not contain any Users
objects that might injure the driver in case of • All buttons should be visible and reachable
a collision. from a comfortable driver possition.
• The driver must have access to a direct arc of • Must not limit the range of view.
visibility ahead. • Should be able to see all buttons at all time.
• A fully harnessed driver must be able to va- • Displays should be angled towards the driver.
cate the car at any time without assistance in • Should have a mounting point for the stering
less than 10 seconds. wheel
• The emergency shutdown system must be
operable from the driver position. Team Goals
• The door opening muste have a minimum • Should be made of carbon fiber, and have a
dimension of 500x800 mm. maximum weight of 2 kg.
• Should cover mechanical and electrical parts
Other Sub-Systems in the front part of the car.
• Must have a 30 mm wide hole to mount the • Should have a simple shape that fits with
steering coloumn in. the rest of the car without drawing to much
• Must be stiff enough to keep the steering attention.
stable. • Must be possible to produce at NTNU by the
• Must include one start/stop button which team.
Figure 3.2.2: An 3D-model with the outlines for the dashboard were given to group
working with idea generation
The students working on the dashboard were using UGS NX for the surface modelling, while the rest of the interior
was designed in SolidWorks
3.2.2.6 Cooperation With Other Students ideas. In this way one could make sure that the
The design process around the dashboard were process moved in the direction that was wanted,
done in cooperation with fourth year students at and that the finished product would fit into the
Institute of production and materials. As a part of car.
the course PuMA 8, where one of the goals was
to learn surface modelling in NX. The final delivery from the PuMa students was
several early concepts that could be used as
The design process were a iterative process a baseline to develop the finished dashboard.
where the team gave the students an outline Because of a short time limit, the last part of the
of what was wanted in the form of 3d-models, design process was done by the designers on the
basic dimensions and rules. In a few weeks the team. The product was taken from early concept
students produced initial sketches and came up phase into a finished concept and production in
with a lot of ideas. In the start the goal was not to a short time, to make sure that no other sub-sys-
limit the students to much, and in this way come tems would be delayed due to the late produc-
up with as many new ideas as possible. tion of the dashboard.
Along the design process, short meetings were

held so the team could give feedback on the

Figure 3.2.3: Concept 1 had a removable top cover to provide easy access.
Figure 3.2.4: Concept 2 consisted of a light construction attached to both the monocoque shell and the floor. It had
no removeable parts.
3.2.2.7 Concept Development easily removable top cover should be attached.

When the PuMa students finished their work, The purpose of this cover would be to hide the
there were two definite favourite concepts. The mechanical parts and wires beneath it.
two concepts were quite different and hard to
combine. One of theme needed to be chosen to Concept 2
continue the further work. Concept 2 was not attach to the wheel wells, but
in the top shell and in the floor. The beam in the
Concept 1 middle of the car could be used to place buttons
The first concept was not that different from the on. It could be a problem when the driver should
one in the DNVFF. It was resting on the wheel evacuate the car. Before this concept could be
wells, and therefore also making the whole car chosen, a lot of testing would have to be done on
stiffer. The front beam was curved and closest the placement of this beam.
to the driver in the middle. It could support the
steering wheel in a best possible way. The curves The second concept would support the steering
on the bottom should provide more than enough wheel in a really good way, but did not hide the
free space for the feet, and to access or evacu- electronic components in the same good way as
ate the car in less then 10 seconds should not the first concept. On the other hand it would give
be a problem. On the top of the dashboard an the whole car a really light expression.

The final design of the dashboard, containing two different parts; the top cover and the front beam.
Figure 3.2.5: 3D-model of front beam mold ready for Finished mold ready for production of the front beam.
milling. The mold is milled in two parts and then glued
together later.
3.2.2.8 Final Concept the the second concept it would also give bet-
After a short meeting the team decided to make ter mounting access beacuse of the removable
a dashboard similar to the first concept, but with cover on the top. This would make the job for the
simplified curves. This way the concept would mechanics and the cybernetics a lot easier. The
have few double curved surfaces and be easier to production would also be a lot easier.
produce in carbon fiber.
The dashboard was produced in two separate
The reason why the first concept was chosen, parts, the front beam and the top cover. The
was mostly because it would provide much free molds were milled out of ebaboard and sanded
space for the feet and simple access. The team down. The carbon fiber was then layed. Three
could not risk that the driver would not be able layers of carbon fiber was used in each part. The
to evacuate the car quick enough, and therefore front beam was reinforced with a Rohacell core.
not pass the technical inspection. Compared to The end result was glued in place inside the car
and then foiled with matt black foil.

Emergency Blinker indicators
shut-down
Windshield wiper
Start/Stop
LED display with control button
3.2.2.9 Placement of Buttons and Display Start/Stop

The buttons were used for several purposes. The Should also be accessable from the outside, and
only common demand was that they should should therefore be placed far to the right.
be accessable from the drivers driving position.
It was therefore important to think thoroughly Windshield wiper
through where they should be placed. This is the least important button, since the wiper
probably never is going to be used. It should
Emergency shutoff be possible to reach it from the outside, just for
Should be places close to the door so it was ac- adjustment purposes, and it should therefore be
cessable both from the inside and the outside. placed on the left side of the dashboard.
LED-screen Indicator lights

Could be hard to read in sunlight and should Can be placed wherever it is most convenient, as
therefore be shielded from the sun. It should long as they are in sight of the driver. Since no
never be covered by the steering wheel, neither buttons would be place on theright side of the
the hand while opperating the control button. steering wheel, this would be a perfect place-
The control button should be placed right beside ment for these lights.
the display. Summing up these arguments, the
best place for the LED-screen would be out to the
right, shielded from the sun by the monocoque.

The front beam of the dashboard was glued on to the wheel wells. The reinforcement with a foam core made it
stiff enough to support the steering wheel.
The finished dashboard with holes for emergency shut-off button, start/stop button, windshield wiper button
and steering rod.
Both the front beam and the top cover weas foiled with matt black foil for a better finish.
The main sponsor logo was also placed on the top cover.
The former steering wheel in the DNVFF was made of carbon fiber wrapped around around a foam core. This,
in addition to the use of really lightweight buttons gave the steering wheel a low weight.
3.2.3 STEERING WHEEL

Since the outcome of the competition was also 3.2.3.1 Last Cars Steering Wheel
depending on the skills of the driver, the design The steering wheel in the former car have been
of the steering wheel could affect the result. changed a few times, and the last version was
Through the design process the main focus was a simple steering wheel of carbon fiber. It was
on ergonomics to increase the performance of made by cutting out a foam core in the desired
the driver. shape, and then wrapping carbon fiber around
it. The result was a steering wheel with a rough

finish. Most of the buttons were placed on the 3.2.3.2 Necessary Functions
dashboard, while the few that were placed on As discussed in the dashboard chapter, the steer-
the steering wheel were small buttons with little ing wheel should contain all the buttons that
respons when pressing them. The buttons were are used during the race, so that the driver never
placed far to the middel, and were therefore not have to let go of the grip. The needed buttons
accessable without changing the grip. would be:
Because of the way it was contructed, the steer- • 6 engine control buttons.
ing wheel had no cover for the wires in the back. • Horn.
This could be an advantage when mounting it, • Indicators (left/right).
but did not look very nice. • High beam.
• Low beam

3.2.3.3 Production Methods Vacuum Formed Plastic
When choosing production method for the This is a fast production method. Plastics that can
steering wheel there were especially two be used are poly carbonate, ABS or PET. One will
important aspects; weight and production time. need a positive mold to vacuum form the plastic.
In the end three different production methons
were considered. Aluminum
This is the heaviest option compared to the
Carbon Fiber others. The product can be made of sheets, or
This is a strong and light material. Pre- milled from a solid block.
impregnated fibers or wet layup can be used.
Carbon fiber can be laid in a negative or a
positive mold. The surface in contact with the
mold will get the smoothest surface.

3.2.3.4 Inspiration To differentiate the ideas and to be inspired to do
The collecting of inspiration could be devided something new, it was decided to collect inspira-
into three parts. Shapes of existing steering tion on different steering solutions. Solutions like
wheels were studied, just to get an idea of how a formula 1,student formula, airplanes, and other
normal steering wheel looks. eco-marathon cars were of course obvious to in-
clude in this collection. In addition solutions like
Testing similar products was the largest source helicopters, trains, normal cars, scooters, boats,
of inspiration. Walking through computer stores bikes and gaming consoles were included.
trying out different joystick- and steering wheel
solutions inspired to a lot of new ideas. Especially
the ergonomics in the product were inspiring,
since they at most times are well designed.

3.2.3.5 List of Requirements • Should include all buttons in one grip.
As in the process of designing the dashboard the • Buttons must give clear feedback when
requirements can be devied into four different pressed.
categories; Rules, users, team goals and other • All on/off buttons should have light indica-
sub-systems. tion.
Rules Team Goals

• Must be possible to use the steering wheel • Should have a nice finish.
with gloves.
• Must not prevent the driver from getting out Other Sub-Systems
of the vehicle. • Must contain 6 engine control buttons.
• Steering must be achieved by using a steer- • Must contain two buttons to control the front
ing wheel or section of a wheel. lights (high beam and low beam)
• Must contain at least one button to control
Users the horn.
• Should be customized to fit the selected • Must contain two turn indicator buttons
driver, but possible to use for everyone else. • Must contain one warning flasher button.

Since the monocoque was not ready for testing when the interior design started, it was
built a mockup to do ergonomic testing and set an outline for the design
Figure 3.2.6: Skething was done to explore ideas, but most of the ideation process was done with physical models
and 3D-models.
3.2.3.6 Idea Generation dimensions, it was hard to relate to the ideas. It

To be able to finish the ideation process as fast as was therefore desided to build a full scale mock-
possible, rapid iterations between fast freehand up of the car.
sketches and physical models were done. The The mock-up had the same dimensions as the
freehand sketches were used to explore ideas, real car, but only the left half was built. It was
while the pysical models were used for testing now easy to get instant feedback from a person
them. Most time was spent on the models, since testing different concepts.
the quality of a steering wheel is highly depend-
ent on the ergonomic properties. This way, a To do the testing of the ideas, foam models were
lot of feasible ideas could be generated over a produced. These models were tested on the
short period of time, and put together into final driver or other persons with the same size. The
concepts. goal was to build a steering wheel that fit the
driver perfectly, and could be used by most other
3.2.3.7 Ergonomic Testing persons.
When the design process for both the steering
wheel and the seat started, there was no finished
car to test them on. Without the possibility to
physically test the ideas and basic ergonomic

Figure 3.2.7: Concept 1
3.2.3.8 Three Different Concepts with the thumb, and the rest with the index fin-
All the ideas from the ideation process were ger. On each side of the steering wheel remove-
grouped into three different concepts, all made able lids will be placed to make the mounting
with different production methods. and wiring simpler. All the wires are pulled from
the buttons, through the tube and the steering
Concept 1 rod, and out behind the dashboard.
The first concept can be produced by cutting out
a foam core in two parts, and put a tube for wires Because of the production method, this would be
through it. Carbon fiber can be wraped around it, a really lightweight product, with quite rough fin-
and holes for easy access cut out. ish. The production is simple, but a bit more time
consuming than the other concept.
The basic idea is that all buttons can be accessed
without changing grip; all the control buttons

Concept 2 made in just a few minutes.

The second concept is vacuum forming either
PET or poly carbonate in two separate parts. Six Like concept number one, this concept also got
screws are used to screw it together to form a all the most important buttons accessable within
steering wheel. All the wires are hidden inside to the same grip. All the engine control functions
give it a clean and organised look. The fact that it are placed in a circle around the thumb. The
is made out of two parts makes both mounting other buttons are also accessable by the thumb,
and possible adjustments a lot easier, since it can exept the indicator lights, which are placed on
be taken apart. the back of the steering wheel.
Vacuum forming is a fast production method,

since it is only the milling of molds that takes
time. A big advantage is that spare parts can be

Concept 3 Since the concept is made out of only one plate,

The third concept is the simplest of the three there is no possibility to mount buttons on the
and also the easiest to produce. The main part is back of the grip. It is hard to place all the buttons
milled out of a aluminum plate, while the grip is within the range of the thumb.
milled out of rubber. All the buttons are screwed
on to the plate, and soldered on the back. The The surface finish of this concept would be better
wires are not hidden in the same way as the than the other concepts. All parts are milled out,
other concepts, but are still pulled through the and the details could be done really well. The
steering rod to make the driver environment production time of this concept would by far be
more organised. the shortest one. On the other hand it would be
the heaviest alternative.

Figure 3.2.10: Criteria matrix
Choosing concept weight and production time was therefore rated

Since the decition of which concept to choose as the most important criterias. The over all finish
was depending on so many different factors, it and easy mounting was of still important, while
was decided to create a criteria matrix to make the price would have almost no impact on the
the process easier. The different criterias were choice.
rated from one to five, where five was the most
important. When rating, number two ended up as a clear
winner.
Good performance in the competion was the
main goal, and the one thing that could really af-
fect that was the ergonomics. This together with

Figure 3.2.11: The molds for the vacuum forming was first designed in SolidWorks and then milled out in a CNC
milling machine
The steering wheel was vacuum formed in two Holes for the buttons were drilled in the plastic
parts
3.2.3.9 Production rial used was 2 mm black PET, with a slightly

Drawings for production of the molds had to be structured surface.
made. Two molds were modelled in solidworks,
with a 2 mm offset of the original shape. 5 de- When the vacuum forming was finished, the
grees of draft angle as well as small indents were holes and indents were drilled out. Then the
also added. buttons could be mounted and details like logos
could be added. Finally all the buttons were
The milling of the molds and the vacuum form- wired and the steering wheel was mounted on
ing was done in the workshop at IPD. The mate- the steering rod.

All the wires from buttons on the steering went through the steering rod which was placed in the middle of the
steering wheel
The finished steering wheel mounted to the dashboard

The seat from the DNVFF placed inside the cockpit of the DNVFF2
3.2.4 SEAT
To be able to performance the best, the driver the finish and ergonomics have generally been
needs to have a comfortable driving position. poor beacuse of an extreme weight focus. Three
The seat is one of the most important factors in different seats have been made for the DNVFF
providing this. over the years. Two of them have had the pos-
sibility for adjustments, while one of them was
3.2.4.1 Former Seats customized to fit only the selected driver. All the
When looking back at former project reports, it seats were made of carbon fiber to make them as
seems the seat have never been prioritized. It has light as possible.
always been one of the last parts to be made, and

3.2.4.2 Inspiration
The seats found in normal cars are usually
not lightweight. To find relevant inspira-
tion, one should rather look at lightweight
chairs, mesh constructions and hammock
solutions. Transferring solutions of chairs
over to different car seat concepts, gave a
lot of inspiration.
-
3.2.4.3 Possible Materials Foam
Different kinds of foam could be used to make
Carbon Fiber the seat more comfortable. It could also be the
As in most of the other parts, carbon fiber can be main part of the seat. The recommended foam
used to produce the seat. Without too many lay- for this application weighs 45 g/l, which is really
ers of fiber, the material can be quite flexible, and light compared to the other materials.
therefore perfect to make a comfortable seat.
Aluminum
Fabric Pipes and frames of aluminum could be an alter-
A lot of lightweight chairs are made out of mesh native to the carbon fiber. It can be easier to work
or lattice which are buckled up between several with, but heavier.
mounting points. This is probably the lightest
solution.

Monocoque
Firewall
Seat belts
Dashboard
Door
Figure 3.2.12: Seat Interfaces
3.2.4.4 Interfaces a comfortable range.
Monocoque/Firewall Door
The monocoque and the firewall are surrounding According to the rules the driver must be able to
the interior, and the seat must be attached to one evacuate the vehicle in less than10 seconds. This
of these. means that the seat must not prevent the driver
from opening the door and getting out in time.
Dashboard/Steering wheel
The seat has to be placed in a way that makes it Seat belt
possible for the driver to reach both the dash- A 5 point seat belt must be mounted, and it has
board and the steering wheel, while still having to fit around the driver sitting in the seat.
a comfortable driving position. Some of the
buttons on the dashboard are rearly used in a
driving situation and does not have to be within

150°
Lumbar support
15-30°
Figure 3.2.13: Basic Ergonomics
3.2.4.5 Basic Ergonomics The parameters were set by doing a lot of test-
The car was already designed and put into ing, first in the car mock-up, and later in the
production when the design of the seat started. finished car. A test person, either the driver or a
The seat had to be designed to make the driver girl at about the same size were used. In addi-
comfortable while driving the car. tion recommended dimensions and angles were
found in the Human Scale collection.
A normal car seat is usually designed to fit the
range of people between 5 percentile women When driving a car in Shell Eco-marathon, it is
and 95 percentile men. It was already known that hard to determine wanted driver position.Most
the driver was 50 percentile woman. Parameters times the decitions were based on the drivers
like the distance from driver to steering wheel, wishes. It was always important to keep a good
perfect arm angle, back angle and where to place balance between a close steering wheel, and one
the lumbar support could be set to exact values further away which would be more comfortable.
instead of a range. It has to be remembered that
the seat should also be usable for other persons.

Must be able to se 90° to the side
Sight straight forward
3.2.4.6 List of Requirements Users

The requirements for the seat were mostly set by • Should be customized to fit a 50 percentile
the driver to make the cockpit environment both woman.
comfortable and user friendly. On the other hand • Must be possible to adjust to fit users outside
the most important requirements were set by the the main user group.
rules. • A 50 percentile women should have a 15-20°
angle in the back and a 25-35° underarm
Rules angle when sitting in the car.
• Must not prevent the driver from evacuating • A 50 percentile women must have lubar sup-
the vehicle in less than 10 seconds. port and support along the spine.
• Must be possible to buckle up the seat belt
Other Sub-Systems when sitting in the car.
• Should be possible to remove from the car in
a fast and simple way. Team Goals
• Should have a nice finish.
• Should weigh less than 500 g.

Figure 3.2.14: A lot of explorative sketching was done to generate as many ideas as possible
3.2.4.7 Idea Generation stages of the ideation consisted of more rapid

The ideation process started with exploring iterations between sketching, physical models
sketches to get to know the product. In addition and 3D-models. This generated a lot of different
a lot of physical mock-ups were built and put into ideas, while at the same time validating them
the car mock-up to test limitations, and decide and speeding up the process.
on basic dimensions.
When both sketches were drawn and physical

mock-ups were built, the result were put
together into more detailed ideas. The later

Figure 3.2.15: 3D-modelling was used to test different ideas and to decide on different dimensions in a fast way
A full scale mock-up was used for testing before the monocoque could be used for this purpose
Fast mock-ups of different concepts were built during the process to validate the diffent ideas
Figure 3.2.16: Idea 1
3.2.4.8 Three Ideas support when tightened. This idea would prob-
The ideation process resulted in a lot of sketches, ably be the lightest solution, but it would not
and to get an overview of these, they were or- be adjustable as it was planned at this stage. If
ganized into three different idea groups. Because the idea had been developed further, adjustable
of the lack of time, these were not finished con- fastening points could have been added, and
cepts but rather more developed ideas. therefore made it more flexible.
Idea 1 A tilted pillow was placed on the floor, attached

The first idea group was based on all the ideas with velcro to make it possible to move it back
around use of stretched mesh. A sheet of light- and forth. The bottom of the pillow would be
weight mesh was stretched between four fasten- made of carbon fiber.
ing points in the monocoque. Two of them were
straps attached in strategic places to give lumbar

Idea 2 back and forth. It has three joints which makes

The second idea group would be the most ad- it possible to adjust the sitting angle, and com-
justable one. The back support is attached to the fortable for most persons. The adjustable pillow
monocoque in two points, both with velcro so consist of carbon fiber plates in the bottom to
that it can be moved. It is made of a carbon fiber stiffen the construction.
frame with a stretched mesh in the middle to
both save weight and make it more comfortable
for the driver. The shape is curved to give some
lumbar support.
The back support is attached to the pillow which

is another adjustable part. It is attached to the
floor with velcro and can therefore be moved

Idea 3 persons want to try the car, and it could also be

This is the simplest solution, and is therefore also a problem when evacuating. On the other hand,
the least adjustable. The pillow is the only part a fixed seat is a safe solution, and few things can
that can be moved. This was to keep the design go wrong during the race.
as simple as possible, and have few loose parts.
The back support is similar to concept 2. It

is made as a carbon fiber frame with mesh
stretched between, and is mounted to the mono-
coque with screws at the top and the bottom.
When made, this seat should be optimized for

the driver. It could be a problems when larger

Figure 3.2.19: All the different seat ideas, where idea number 2 was selected for further development.
3.2.4.9 Selecting Ideas most logical choice. It was light, had adjustments
It was hard to choose the final concept. Discuss- and would be easy to produce. This also needed
ing with team members and other designers, it to be developed further, but this was not con-
was early agreed that idea number 3 would prob- cider beeing a time consuming job.
ably create problems during the technical inspec-
tion. Both idea 1 and 2 were very interesting. Idea
1 was really lightweight, while idea 2 had all the
adjustment possibilities.
It was agreed that if idea number 1 should be

used, it had to be further developed to include
adjustment possibilities. Because of the lack of
time, this was not a possibility, and idea 2 was the

Figure 3.2.20: Details of the final concept , cushions added
3.2.4.10 Detailing of the Final Concept there was no time left in the end to produce this
To make a final concept the basic ideas from idea mechanism.
2 were taken further. A lot of changes were made.
The angle adjustments in the pillow were re-
The back support got a totally new and more moved. It was found unnecessary during testing,
exciting shape. Stretched mesh was replaced by and it would add extra weight to the car. The final
thin pillows along the spine supports and across pillow was milled out of lightweight foam, and
the lumbar support. attached to the floor with velcro.
To make the concept even more adjustable, a

width adjustment was added. This was done
because of the discoveries made during the final
mockup testing of the concept. Unfortunately

The mold was milled out of Ebaboard with a CNC-milling machine
3.2.4.11 Production glossy paint. Small strips of lightweight foam

The molds for the back support were modelled in were glued on along the spine and on the lumbar
SolidWorks as negative molds. This way the part support. The back support and the pillow were
would get a smooth surface without too much attached to the car with velcro.
sanding. When the 3D-model was finished, the
molds were milled out in ebaboard and sanded On each side and in the front middle of the seat,
down. The carbon fiber layup was done as a wet a 5-point seat belt was attached. The belt was the
layup with a core of rohacell. To give the driver a smallest found, originally meant for junior rac-
more comfortable position, the layup was made ing. It was modified with some lighter parts, and
to give a small flex in the back. adjusted for SEM rules and use. The weight was
reduced from 2.5 kg to 1.5 kg.
The finished carbon fiber parts were cut out,
sanded down and painted with transparent

The mirrors of the DNVFF were larger than required by the SEM rules
3.2.5 REAR VIEW MIRRORS

According to SEM rules, the car must have two with benefit be reduced. The mirrors are placed
rear view mirrors. Since this was one of the less inside the cockpit to reduce the aerodynamic
important featurers affecting the performance of drag.
the car, the design process was relatively short.
3.2.5.1 Mirrors on DNVFF

Same mirrors have been used on DNVFF since
2009. The mirror surface is several times larger
than what is required by the rules, and could

3.2.5.2 Inspiration
A lot of work is often put into design and produc-
tion of beautiful rear-view mirrors by the car pro-
ducers. It is therefore no problem to find a lot of
good inspiration on this area. On the other hand
very few car producers are focusing on weight
and aerodyncamics when designing the mirror.
Espesially indoor mirrors are a rare thing to find.
Figure 3.2.21: Detailed model of the interior without mirrors
3.2.5.3 List of Requirements Users

Since the rear-view mirrors have few interfaces • The mirrors should be adjustable in a way
with other sub-systems, the list of requirements that makes it possible for the selected driver
is based on the rules, and on requirements from to get the perfect view.
the users. • The quality of the the mirror glass should be
good enought to spot competitors driving
Rules behind the car.
• The vehicle must be equipped with a rear- • The mirrors should not block the drivers view.
view mirror on each side, with a minimum • If placed on the outside, they should be easy
surface area of 25 cm2. to clean.

Figure 3.2.22: Explorative sketching was done to generate ideas
To get an idea of how big the mirrors should be, different cardboard models were made with the required area. By
using these models together with the sketches it was possible to make 3D-models of the ideas
3.2.5.4 Idea Generation two different categories; interior and exterior

Quick drawn freehand sketches was the main mirrors. The exterior mirrors were the biggest
tool in the ideation process of the mirrors, but challange. They had to be aerodynamic, really
also physical models were used to get an impres- small and have nice details.
sion of the actual size . When experimenting with
pysical mockups we realised that 25 cm2 was Most interior sketches were of mirrors hidden
really small, and did therefore make different behind the a-beam, which was the only place
samples of this size in different shapes. This was where they did not block the drivers view. The
really handy through the rest of the process. interior was easier to draw, since the main focus
there should be to hide them, not give them nice
The resulting ideas can basically be divided into details, and a beautiful shape.

Concept 2
Concept 3
Concept 1
Figure 3.2.23: The three different concepts seen from the drivers view. Closeup of the different concepts are shown
below
Concept 1 Concept 2 Concept 3
3.2.5.5 Different Concepts small, they can be hard to use when mounted on
The result from the ideation process was devel- the outside of the car.
oped into three different concepts. All of them
were really small (25-30 cm2). One was exterior, Concept 2
while two were interior. Was the first interior concept. It had a long shape
so it could be placed on the a-beam. With joints
Concept 1 in each end it could be adjusted to fit the driver.
The only exterior concept, and it has a flat shape When testing the mirrors later in the process it
which gives it a light and floating look. The shape was discovered that it could be hard to spot cars,
is made as aerodynamic as possible, but will because it was too narrow.
always induce some drag. Since the mirrors are

Figure 3.2.24: The main part for the mirrors designed in SolidWorks. The model was planned in a way that made it
easy to mill
The finished mirrors seen both before and after beeing mounted in the car
Concept 3 also count in a positive way.

An interior mirror with a triangular shape. It can The mirror base was milled out of acryllic, and the
also be placed on the a-beam. The mirror is at- glass and brass were glued on. It was mounted in
tached with a sheet of brass, this way it can be the car while the driver was sitting in the driving
bendt and adjusted many times without braking. position.
3.2.5.6 Production and Finished Product The finished product is an anonymus mirror, that
Concept 3 was the chosen concept. It would is almost invisible from the outside. It has a satis-
create the best view and not interfere with the fying view range to the back, and does not block
aerodynamics. The fact that the adjustment the front view.
mechanismen would be so easy to produce, did

3.3
SUSPENSION
3.3 SUSPENSION
3.3.1 FRONT SUSPENSION the attached wheel. The placement of these rela-
tive to each other are absolutely crucial to obtain
It was important to keep all wheels in contact the desired motion.
with the ground at all times for the sake of safety
and fuel efficiency. 1. Top swivel joint
2. Lower swivel joint
The aerodynamics, size and weight of the car 3. Steering joint
has been prevalent during the design phase and 4. Axle/hub
overruled some of the design freedoms of the
suspension. For instance, the diffuser underneath
DNVFF2 is crucial for eliminating aerodynamic
lift (Endresen, et al., 2011) and its geometry af-
fects where the lowest mounting points for the
suspension linkages can be placed. This means
that an optimal roll center, placed as low as pos- Figure 3.3.1: Typical steering knuckle concept
sible, cannot be achieved as the linkages will
slant upwards into the body, especially for the Change in camber angle
rear suspension where the diffuser slants up and When the car drives through a curve, the lateral
away from the ground as it meets the car’s tail. acceleration will transfer weight to the outer
wheels. With the ground as reference frame,
Figure 2.3.2 shows the parts that the steering the car will tilt to the side, and depending on
knuckle must have in order to provide proper the suspension kinematics, the wheels will alter
steering and spring action. Swivel joints allow the camber angle. Since the camber angle both af-
knuckle to rotate about its vertical axis to steer fects energy efficiency and lateral thrust force,
the camber angle relative to the ground should

Figure 3.3.2: Double wishbone system, parts annotated
remain fairly constant, or rather it should at least With the car as reference frame, the negative
not become positive. camber change can lead to unwanted effects
regarding suspension.
With a positive camber angle less lateral force
than required may be developed, with the The chosen solution was to optimize the camber
possibility of slipping. The cornering stability angle change for minimizing the ground-to-cam-
is reduced because the contact point is moved ber-angle during cornering action. With an axle
towards the center of gravity. width of 100 cm, and a maximum delta change of
±2 cm, the car’s roll angle and the corresponding
A negative camber angle would rather increase
the stability, but could also lead to over steer.

Figure 3.3.3: Camber angle change during spring action
camber angle compensation can be calculated as

follows:
The CAD assembly with its constraints was used

to modify the link geometry until the desired
camber angle change was achieved. A model
with three sheets spaced 2 cm apart was made to º.
show the different wheel positions during spring Figure 3.3.4: Top view front view of front left suspen-
action when placed correctly beneath the shell sion links
model of the car. From left to right in figure 3.3.3:
Lowest (wheel hanging), ride height, highest
(compressed by bump or hard braking or corner-
ing).
Wishbone Angles
Seen from above, the links form triangles that,
coupled with the rod ends and clevis mount
points as hinges, define the vertical swinging
motion of the knuckles, and otherwise complete-
ly lock the knuckles from moving sideways or Figure 3.3.5: Viewed from beneath, front right sus-
forward or backwards. The angle subtended by pensions
the links in the horizontal plane determines the
force absorbed during braking and cornering.
The angles can be optimized for reducing stress
during braking or cornering. The middle ground
is 90º. However, to avoid the links interfering with
the wheels turning about the steer axis, and
because the car will corner harder and more
frequently than braking, this angle was reduced
to 80º.
Figure 3.3.6: Rear view of front left suspension links
Figure XX - Camber angle change during spring action

For the lower links, the rear link had to be rotated
and moved forwards to give space for the tie rods This greatly simplifies the derivation of the con-
Since energy efficiency and safety are so impor- straint equations to be solved as a set of matrix
tant, the no bump steer-requirement required equations. The steering knuckle can be modeled
that the tie rods be moved down into the plane as a beam supported on two points where TB is
of the lower links. The angle of the lower, rear link the torque generated from the braking force
was thus reduced to 8º from the lateral axis of the PB = P2, RW is the wheel radius and L is the height
car. of the knuckle:
Calculation of Link Forces

The principal forces acting on the wheels that the
suspension must support are derived and read-
ily explained in the Project Report autumn 2011
(Endresen, et al., 2011).
The P1 and P2 forces will be used to determine

the forces acting through the primary linkages,
and P4 the force acting through the coilovers on
the front suspension. To find these forces, the
system(s) can be modeled as rigid bodies with
loads and fixed constraints where the bodies are
connected to other linkages or mounting points
on the body.
Figure 3.3.8: Force and constraints on the knuckle
Figure 3.3.7: Rigid body model of front suspension,

where PB and PC are the braking and cornering
forces, and C1, 2, 3, 4 are the support forces acting
through the wishbone links
Figure 3.3.9: Simplified force and torque on knuckle
The calculations are simplified by splitting the
mechanical joints into three separate problems: To further simplify the system, the reaction forces
can be annotated so that the two unknowns for
1. The steering knuckle experience longitudinal the knuckle are clearly shown in figure 3.3.10.
braking force or lateral cornering force
2. The top linkage triangles
3. The lower linkage triangles

Figure 3.3.10: Steering knuckle modeled as a beam
with load and reaction forces
This setup can also be reused when the effects of Figure 3.3.11: Constraint model for top arms during
the cornering force are calculated. The system is braking
then expressed as the following linear equations:
In matrix form this gives:
Thus, given that PB = P2 = 550 N, RW = 279 mm

and L/2 = 100 mm = 0.1 m:
TB = PB RW = 550 N × 279 mm = 154 Nm
Figure 3.3.12: Constraint model for lower arms dur-

ing braking
The top arms must thus exert a backwards force Solving for the support forces in the beams can
of 495 N and the lower arms a forwards force of be done with the followingg equation:
1060 N. To decompose these reaction forces into
the arising axial forces in the top and lower links,
the following model and equations are derived:
Where P is the force acting on the swivel joint, Cu

and Cv are the axial support forces in the beams,
and θ1 and θ2 are the angles from the horizontal
plane. Applied to the top arms this becomes:

And applied to the lower arms this becomes:
Here, a positive value means that the element

undergoes tensile load, and negative values com-
pressive loads.
For the cornering force, we observe that we can

scale the result for the steering knuckle and avoid
Figure 3.3.14: Constraint model for lower arms dur-
solving the same equations again:
ing cornering
The reaction forces on the steering knuckle

become:
For cornering in the opposite direction, the forces

are inverted.
From these calculations we observe that the

greatest axial loads in the links occur during brak-
ing, and the force of greatest magnitude (1392
N equals 1.4 kN) will be used as dimensioning
criteria (with a safety factor) as it simplifies CAD
modeling and manufacturing by reusing parts as
much as possible.
Figure 3.3.13: Constraint model for top arms during

cornering

Calculation of Coilover Forces Since the ground-tire force is higher (about
71% of the car’s weight, 971 N, compared to the
cornering force of 510 N in section 3.3.1.4) than
the cornering force, this will be the determining
scenario for the tie rods.
The force on the tie rod occurs from the transfer
of the force on the ground acting through the
king pin axis to produce torque. The “Caster-
g is:
length”
The tie rod force follows from the torque:
As the wheel turns about the steering axis, the tie

rod force will increase.
Figure 3.3.15: Diagram of weight force and link Anti - Dive
geometry view from behind the car, where PW is the The dive effect of the weight transfer towards the
weight on the wheel and C5 is the coilover’s reaction front when braking can be countered by exploit-
force ing the generated torque from the brakes to
generate torque on the body that tries to lift the
Since cos 9° equals1 the calculation for the rela- front up from the dive (Gillespie, 1992). Experi-
tion between the force Pw and C5 becomes: mentation with the top arm plane angle revealed
that it affected the caster angle during spring
action, which again influenced the steering-angle
And with: of the wheel, which again could adversely affect
handling and fuel efficiency. The caster angle
changes as a result of the top swivel joint moving
forward or backwards, so that the tie rods change
angle in the horizontal plane. Therefore, the
The negative sign indicates that the coilover
anti-dive was reduced so that the caster change
undergoes a compressive load, or if it’s unit vec-
was negligible, while still trying to maintain some
tor points up to the right in figure 3.3.15, it exerts
degree of anti-dive effect.
a reaction force down to the left. As the wheel
bounces up and down, the angles change and
the multiplying factor will change too, however
this value is useful for the coilover calculations.
Calculation of Tie-Rod Forces

One of the load scenarios, which happened dur-
ing the competition in 2011, is that the entire
weight of the car lands on one of the front
wheels during a less fortunate lift. The car is
assumed to have a roll angle of 45° if this event
should reoccur. To calculate the force experi-
enced by the responsible tie rod, the following
criteria are set:
1. Car roll angle is 45°. The weight force is then

Figure 3.3.16: Sketch setup in NX to obtain measure-
multiplied by sin 45° = 1/21/2
ments for anti-dive calculations
2. The caster angle is approximately 8°
3. The tie rod length is 72 mm

By splitting up into torques arising from the
lower and upper arms, this becomes:
Figure 3.3.17: Force diagram of dive/anti-dive con-

tributing forces From this it can be concluded that the non-hori-
zontal plane of the top arms significantly reduce
The knuckle reaction forces arising from braking, the dive effect; but no more than a quarter.
calculated in the previous section, can be used Although the anti-dive angle had to be reduced
to estimate the resulting dive effect. The center to avoid “bump steer”, some effect positively
of gravity of the car is assumed to be fairly low, remains, and the design choice is verifiable.
namely 50 cm above the ground. Disregarding Unfortunately, it will not be possible to measure
the anti-dive effect completely, the dive-torque the effect of the anti-dive geometry once the test
from the front wheels braking is: driving starts, as the mounting points cannot be
moved.

Figure 3.3.18: Front and back view of the hub
3.3.1.1 Hubs FEM simulations were only done for braking and
The hubs were made for easy installation of bear- cornering, since the car’s weight is transferred
ings, held in place by retaining rings. directly from the rims, through the hub, to the
bearings.
Figure 3.3.20: FEM analysis of hub during braking
The simulation of the braking situation shows

that the stress concentrations stay below the fa-
tigue limit of Alumec 89, which is about 100 MPa.
Figure 3.3.19: View of hub assembly
The hub design sports a double set of lug bolt

holes and brake disc holes. These extra sets of
holes can be used as spares in case the threads
should fail from wear or other accidents. The five
lug bolts concept was chosen over the previous
center nut solution as it allows faster machin- Figure 3.4.21: FEM analysis of hub during cornering
ing of the hubs, and no slow machining of large
center nuts. During cornering, the hubs undergo much lower
stresses.
On recommendation from SKF, the internal hole Three hubs were produced by NOMEK AS for
tolerance was set to JS7 to allow a slight interfer- about 6 700 NOK, including the material, Alumec
ence fit with the ball bearings to allow easy inser- 89.
tion and replacement.

Hubs and coilover parts received from Nomek
3.3.1.2 Axles
If the axles deform during operation they might
interfere with the brakes. Sufficient stiffness is
therefore required.
Since only the outer circumference of the cross

section of a round bar significantly contributes
to the stiffness, the axles could be made hollow
to reduce weight without significantly reducing
stiffness.
Figure 3.3.23: FEM analysis of axle during braking
The axle design was analyzed under the hardest

braking scenario. Even though the maximum
stress concentration climbs above the fatigue
limit of aluminum, the ball bearings and spacers
will help transfer loads to the face of the conic
section, so the real stress concentrations are
expected to be much lower. The deformation was
Figure 3.3.22: CAD model of axle so low (less than 0.1°) so the brake discs will not
be affected.
The axles are improved from the previous ones
by using a large, conic flange to transfer the The axles were produced from a round bar of
bending moment from the car’s weight and brak- Alumec 89.
ing to the knuckles.

3.3.1.3 Steering Knuckles tion.
The use of carbon fiber was recommended by
HPC, and would allow low weight. Carbon fiber
composite parts are often shell-shaped and differ
from the more massive solids usually made in
metal, due to the anisotropic properties of com-
posites. However, the steering knuckles will still
require a rather massive piece.
One of the strong arguments for massive carbon

fiber was that less material would go to waste in
the production. If the knuckles were to be made
in aluminum, they could either be sintered (ad-
ditive) or machined from a solid block; both of Figure 3.3.24: Ackerman steering based on sketch in
which are expensive, and the latter leaves a lot the PD-journal, wheels and angles annotated. α is
of expensive waste material. Casting aluminum the typical toe-angle when driving straight ahead.
would raise the cost even higher because of the
molds required. The carbon fiber would rather This mode of analysis has two goals:
be cast on a rough mould, and then later be
machined into the final shape (figure 3.3.32). As 1. Ensure satisfactory ackerman steering.
seen in table 3.3.1 this is where the carbon fiber 2. Ensure a turning radius of 6 m (as required by
solution gains advantage over metal. Shell Eco-marathon) or less .
Ackerman Steering The exact requirements for the steering angles

Ackerman steering is necessary to accommo- was not well documented in the literature stud-
date the smaller turning radius that the inner ied. The Ackerman steering geometry is only op-
wheel experience. The behavior has been closely timal at low speeds when the tires develop little
checked in the CAD models. Although the for- to no lateral forces. At higher speeds the behav-
mulas for calculating the angle of the toes can ior changes, the turning center moves forward
be easily derived, experiments showed that this (Gillespie, 1992).
does not necessarily lead to a satisfactory solu-
According to IMechE, (1993) full ackerman is not
Aluminum Massive carbon fiber
Criteria Weight Score Weighted score Score Weighted score
Low weight
5 4 20 5 25
Adjustability
4 1 4 1 4
Manufacturability
3 4 12 4 12
Reliability
4 5 20 4 16
Low cost
4 2 8 3 12
Maintainability
3 4 12 3 9
Weighted sum
76 76
Table 3.3.1: Trade-off matrix for steering knuckle material selection

Figure 3.3.25: View of steering radius check
necessary: They discovered that in a Ferrari F40 the sharpest, and 20 m in most other corners. The
only 40% of theoretical full ackerman was used. ackerman test model was thus used to optimize
the position of the toes on the steering knuck-
With purely mechanical passive steering it is les to achieve optimal ackerman on these curve
not possible to achieve full ackerman steering radii.
throughout all wanted turning radii. A test setup
in NX showed that the steering appears fully ack- As seen in figure 3.3.25, the grid marks the posi-
erman compliant the first few degrees, and then tion of the different turning radii from 2 to 25 m,
the difference increases progressively until the in- with the most important ones annotated. The
ner wheel massively over-steers. This over steer of red bar is an extension of the right axle, and the
the inner wheel is thought to be good, since the white of the left axle, so that the turning center
lateral acceleration of the car (at higher speeds) can be observed as the drag link is moved in the
transfers weight to the outer wheels, but as there CAD assembly, exploiting the constraints on all
will be some scrubbing it may waste energy. Still, the suspension parts.
given that we have been provided with a map of
the race track with annotations of the corners, we
know that the corners are no less than 15 m at
Figure 3.3.26: Verification of ackerman steering for 20 m turning radius

Figure 3.3.26 demonstrates the goal of satisfacto-
ry ackerman steering for cornering radii common
on the race track. The testing showed that, as
expected, the angles can be tuned by modifying
the length of the drag link and tie rods, and the
angle of the toes. Figure 3.3.27 and figure 3.3.28,
shows the verification for smaller turning radii.
Figure 3.3.30: FEM analysis of knuckle during cor-

nering
The stresses revealed by these simulations are far

below the strength of the DB 420 carbon fiber.
Figure 3.3.27: Verification of 6 m turning radius HPC provided 15 mm thick samples of massive
requirement carbon fiber for testing. The tests revealed that
the bolt connections would require reinforce-
ment to avoid delamination. This reinforcement
was made using unidirectional carbon fiber
wrapping as illustrated in figure 3.3.31. Tests
showed that a bolted connection could with-
stand as much as 9 kN, far exceeding the required
780 N.
Production of Steering Knuckles

The knuckles were to be produced from the same
material as the monocoque. Exact properties of
Figure 3.3.28: Finding the smallest possible turning this material were unfortunately not known; the
radius, tilted view. theoretical properties are listed in Appendix B.
Strength HPC suggested that they would cast and cure

The steering knuckles would be hard to analyze blocks of carbon fiber shaped by a mold, which
using shells and laminates. Instead, an isotropic would then be milled into the final shape. Figure
model was analyzed for braking and cornering 3.3.32 explains this production technique. The
scenarios where the results can give an indication mold was machined from Ebaboard 60 in the Pro-
of the forces that must be absorbed by the mas- totype lab and sent to HPC.
sive laminate.
Milling of Carbon Fiber
Based on Børge Holen’s advise; diamond-coated
end mills were used for machining the massive
carbon fiber into the final shape of the steering
knuckles. They were also used to mill out the
monocoque’s fire wall.
The milling was scheduled to be done by Bjarne

Stolpnessæter with the CNC mill in the Prototype
lab. In addition to proper vacuuming and safety
glasses, a set of double-layer coal-filter respira-
Figure 3.3.29: FEM analysis of knuckle during brak- tors were purchased from Clas Ohlson to protect
ing the persons working with the machining. Protec-
tive gloves also had to be used.

Figure 3.3.31: Unidirectional carbon fiber wrapping
for reinforcement of bolted connections
Figure 3.3.32: Mold for carbon fiber knuckles

3.3.1.4 Linkages Thread size M6 (mm) M8 (mm)
The easiest solution for linkages involved using
d2
standard parts. With the sponsorship agree- 5.212 7.212
ment with SKF, and leftovers from earlier project d3
teams, the required number of rod ends could be 4.596 6.596
afforded. Similar types of rod ends are also com- ds
monly used on real cars. 4.904 6.904
Variations of male and female and mixed rod

ends on the different links were considered. A Table 3.3.2: Table of minimum pitch and minor
link with external threads is the easiest to pro- diameters of ISO M6 and M8
duce (figure 3.3.33) and therefore this approach
was selected. The links’ adjustability is obtained
by using left hand threads in one end and right
hand threads in the other. SKF offers rod ends
with both right- and left-handed threads.
For wrought Aluminum 6082 T6, the fatigue

strength is minimum 90 MPa (Granta Design Ltd.
CES EduPack 2011). The fatigue limit of the links
g M6 and M8 are thus:
using
This exceeds the max link force (1.4 kN for links,

Figure 3.3.33: Link consisting of rod ends and nuts 0.53 kN for tie rods) by factors of 2.43 and 3.2,
respectively. The safety factor criteria has thus
Link Strength been satisfied.
The link rods are easily machined from round
bars, and 24 meters of Ø10 mm Aluminum 6082 One other failure mode to be considered is buck-
T6 were purchased from Smith Staal AS. It was ling. The Euler buckling criterion for beams with
decided to use Ø8 mm for the primary links and plain joints that don’t exert bending moment is
Ø6 mm for the tie rods. The strength of the rods given by:
is determined by the strain diameter from the
thread size:
where L is the length, and E and I are Young’s

Where d2 and d3 are the pitch and minor diameter modulus and area moment of inertia. The modu-
of the threads (Härkegård). lus of elasticity for aluminium 6082 T6 is 70 GPa,
The longest link is 116 mm and the tie rods are
For external M8 and M6 these are (Fastener 108 mm long. Thus, the area moment of inertia
Thread Designations and Definitions Pitch, Minor, for the tie rods and the longest
g link mm rods are:
Major Diameters):

These results were within the requirements, and
the fatigue was the dimensioning criteria.
Maintenance-free rod ends with “steel/sinter

bronze composite” were chosen for their accu-
racy and ability to run without lubrication. The
other alternative was rod ends requiring main-
tenance. These have some play between the rod
head and the ball which is undesirable. Analyses
of the suspension setup show that as little as 0.5
mm elongation or contraction of a link affects the
wheel angles.
The rod ends with 8 mm bore (SI 8 C), easily sup-

port up to 5.85 kN, and 3.6 kN for 6 mm bore rod Figure 3.3.35: Solid lower wishbone
ends (SI 6 C), which exceed the required loads
calculated the previous sections.
Even though the double-wishbone concept was

chosen for both the front and rear suspension,
the end result was a variant of multi-link suspen-
sion. Only the lower swivel joints need a rigid
part to connect the coilover, so the top swivel
joint was made easier to manufacture by splitting
it into two coaxial rod end joints located next to
each other. The effect of two off-plane joints was
found to be negligible and thus carried forward.
Figure 3.3.36: Connector and link rods
The lower control arms of the suspension can

be made adjustable and lightweight by using
a small “wishbone connector” that rigidly con-
nects the lower swivel joint, the lower links and
coilover.
Figure 3.3.34: Single rod end on bottom, double rod

ends on top
Wishbone Connectors
For the lower control arms, two solutions were
considered. While a single-piece solid wishbone
(figure 3.4.35) is simpler, it is not adjustable like
the connector with rod links (figure 3.4.36). The
connector solution is also easier to manufacture.
Figure 3.3.37: 3D view of the left wishbone connec-
tor

Figure 3.3.38: Left suspension with connector and coilover
The wishbones for the left and right sides are With an expected vertical load of 780 N it can
symmetrical. be verified that 10 mm bore rod ends should be
chosen over the other.
Rod ends from SKF are primarily meant for radial
loads (relative to the eye). The size of the lower
rod ends was decided to be Ø10 mm.
The permissible load for rod ends is given by SKF

as:
where
• C0 = static load rating [kN]

• b2 = temperature factor (= 1.0 for tempera- Figure 3.3.39: FEM analysis of wishbone connector
tures below 120 °C) during braking
• b6 = load type factor (= 1 for constant, 0.5 for
alternating) Figure 3.3.39 shows the FEA of a connector
during the hardest possible braking. The stress
Additionally, SKF states that “the load portion act- concentrations approach 100 MPa, which is at the
ing perpendicular to the direction of the shank axis limit of aluminum. Since this load scenario rarely
should never exceed the value of 0.1 C0.” (SKF) Thus will occur, the design is approved.
the weight carrying capacity of SI 8 C and SI 10 C
bearings are given by: The wishbone connectors were milled from
Alumec 89 with the Makino CNC machine in IPM’s
workshop .

Milling of connectors in the Makino milling machine
Verification of tool path using graph plot on the Fanuc computer

Tie Rods 3.3.1.5 Mounting Points
The chosen solution was to use the same type Each wheel requires five mounting points: Four
of rods as elsewhere on the suspension, only for the links controlling the spring action mo-
smaller. The Ø6 mm rods with right-/left-hand tion, one for the coilover, and one extra for the tie
threads in either end allowed for adjustability rods. On the rear suspension the tie rods are con-
and unobstructed spring- and steering action of nected directly to a sixth mounting point in the
the front wheels. wheel well, while on the front they are connected
to the drag link. The wheel wells were slanted to
optimize for camber angle changes. HPC recom-
mended to reshape the wells into something
similar of the strut towers on real cars, to attach
the coilovers.
An indent for the drag link was incorporated in

the wheel well. To avoid bump steer while driving
the drag-link ti rod-joint must lay in plane with
the other mounting points.
Figure 3.3.40: Tie rod assembly
Roll- and Strut Bar

A roll bar was considered, but this would increase
the weight (which was already a critical issue),
and add a more complex joint mechanism for
retaining the roll bar in the wheel wells made
us avoid this feature. As roll bars affect the over/
under steer factor, this is an option we retain as
a backup solution in case the car should severely
under- or over-steer.
Using a strut bar was also considered, but the Figure 3.3.42: Front left wheel well on the final body
stiffness of the monocoque was so high that this shell
would be unnecessary weight.
Figure 3.3.41: Left tie rod on suspension seen from beneath

Polymer wedges was produced to attach the
clevises with the right angle.
Figure 3.3.45: Clevises attached to wheel well using

wedges and bushings
A casting resin, RenCast CW 2215 and hard-

ener, with compressive strength 80-90 MPa was
purchased from Lindberg & Lund. (Casting Resin
Figure 3.3.43: The final clevis design. Purple high- RenCast). To reduce vibrations dampers were pro-
lights the vibration damping bushings duced. For the vibration dampers, Flexane 60 was
also purchased. The wedges and vibration damp-
Clevises were used to connect the suspension to ers had to be cast in molds from Ebaboard 60.
the wheel well. These were designed to be light-
weight, easily machined and allow maximum rod 3.3.1.6 Conclusion
end tilt. The design of the front suspension has been
very successful, but the final weight exceeds
The bolt connecting the clevis to the body will be the weight goal slightly. Other strengths of the
glued into the clevis. suspension, such as adjustability, spring/damper
mechanism, high strength and vibration damp-
ing makes this an acceptable sacrifice.
Figure 3.3.44: FEM analysis of clevis
Stress analysis shows that the maximum stress

concentration is just 61 MPa, which gives a safety
factor of 1.6 with regards to the fatigue limit.

Upper A-arm rods
Knuckle
Coilover assembly
Brake caliper
Hub assembly
Lower A-arm rods
Connector
Lower mounting point
Figure 3.3.46: 3D CAD model of the rear suspension developed during the spring 2012
3.3.2 REAR SUSPENSION The knuckle connects the axle, the hub and the
wheel to the rest of the suspension. The forces
3.3.2.1 Introduction acting on the wheel will be transmitted through
The most suitable solution for the rear suspen- the knuckle, to the rest of the suspension and
sion was the double wishbone. This would pro- to the monocoque. It needs to be designed to
vide light weight, proper adjustability and a de- withstand the forces acting upon it, and provide
sirable change of camber angle during cornering. sufficient ride qualities such as suitable camber
A suggestion for a fully damped rear suspension change and optimal roll center height.
was presented in the project period (Endresen, et
al., 2011), and was further developed during the Geometric Optimization
Eco-marathon master project in the spring 2012. The design developed during the project phase
had a problem with the angular displacement of
3.3.2.2 Knuckle the rod ends being exceeded (Endresen, et al.,
2011). This was due to changes done in the
geometry of the monocoque at a late stage of
the project phase.
ΔZ - rear
ΔZ - front
Figure 3.3.48: High vertical distance between the

ground and the bottom of the rear wheel well wall
Figure 3.3.47: Knuckle

Upper mounting point
Clamps
Clevis assembly
Axle
Brake disc
Toe link rod
Toe link extension

Toe link mounting point
optimizing the design of the knuckle.
The 2D-model was used to investigate the effect

changes in the knuckle design would have on
the cars roll center. The roll center affects the
behavior of both the sprung- and the unsprung
mass, and thus directly influences cornering. It is
defined as a point in the transverse vertical plane
through the wheel centers at which lateral forces
may be applied to the sprung mass without
producing suspension roll. The roll center is an
instantaneous point which will move during
cornering. The procedure for determining the roll
center of a symmetric independent suspension is
described below. See also figure 3.3.50.
Figure 3.3.49: Distance between the lower rod ends • Find the virtual reaction point of the suspen-
and the brake disc increased on the new design sion links (A)
• Draw a line from the tires contact patch with
The geometric design of the knuckle was the ground (C) to the virtual reaction point
changed, tilting it 23° inwards at the bottom. This • The roll center (R) is located where this line
also increased the distance between the lower crosses the centerline of the body.
rod ends and the brake disc which was too short (Gillespie, 1992)
on the old design.
The rear roll center is typically positioned 200-
A 2D-model of the rear suspensions front-to-back 250 mm above the ground on a race car. On the
view was made (figure 3.3.52). The geometry of DNVFF2 the rear roll center height is 293 mm
the suspension as a whole was included when when the suspension is in ride height, and 256
mm when it is fully compressed.
Centerline
C
Figure 3.3.50: How to determine the roll center of an
independent suspension Figure 3.3.51: Mounting points for A-arm connector
and toe link on the same horizontal line to prevent
The geometry of the knuckle will also affect the bump- and roll steer
bump- and roll steer, which are small changes
in the suspension’s toe angle when the wheel is
moved relative to the body in bump and droop.
Bump- and roll steer causes poor straight-line
stability, unpleasant vehicle behavior and high
energy losses on lighter cars (Dixon, 1996). The change of camber angle for the rear suspen-
sion is 2.1° during compression. A larger camber
The change in camber angle during cornering is angle change would lead to too much lateral
configured based on the wheels maximum movement between the tire and the ground.
vertical travel of ±20 mm. Approximately it Then the wheel would be outside of the cars
should be equal to the body roll to ensure good body during spring compression, violating the
handling abilities. SEM rules.
Figure 3.3.52: 2D-model of the suspensions front-to-back geometry in UGS NX 7.5. The most important dimensions
are shown in the illustration.
Weight Optimization and Strength Analysis rect distance in relation to the knuckle. To apply
The knuckle was made in massive carbon fiber constraints and forces as close to the reality as
to save weight. A proper strength analysis of possible was emphasized, but the reader must be
the knuckle was done in UGS NX 7.5 Advanced aware that some compromises have been made.
Simulation with estimated material properties The mounting points for the brake caliper have
for the carbon fiber Appendix B. The reader must been neglected to simplify the simulations. This
be aware that the uncertainties regarding the is assumed to have little influence on the final re-
material properties may have lead to small errors sult. The forces applied are based on the calcula-
in the result. tions done in the project report from the autumn
2011 (Endresen, et al., 2011).
1D RBE2 beam elements were used to make
models of the axle and the wheel. This was done
to apply the loads and constraints in the cor-
A
B
y x
F3
F1
F2
Figure 3.3.53: FE analysis setup for the knuckle Figure 3.3.54: Test result from the final analysis of
the knuckle
A B C Value Unit
DOF 1 1 1 1 F1 510 [N]
DOF 2 1 0 1
F2 550 [N]
DOF 3 1 0 0
DOF 4 0 0 0 F3 960 [N]
DOF 5 0 0 0 σmax 233 [MPa]
DOF 6 0 0 0
Table 3.3.3: FE analysis set up for the knuckle. DOF Table 3.3.4: FE analysis set up and result for the
1-3 is displacement in x-, y-, and z-direction, respec- knuckle
tively. DOF 4-6 is rotation about x-, y-, and z-axis,
respectively

Figure 3.3.55: STEP - model of the knuckle imported into NX IDEAS 5.0. The couloured lines show the milling path
Several simulations were carried out and im-

provements were done until acceptable stress-
and strain values were achieved. The final design
consisted of 32 layers of carbon fiber, giving an
overall thickness of 16 mm.
Production
A mold for producing the knuckle was milled out
from ebaboard at the IPM prototype lab. Bjarne
Stolpnessæter was very helpful during this pro- Figure 3.4.56: 3D model of the mold and the
cess. The mold was sanded down and painted knuckle
with polyurethane paint to ensure a smooth
surface. The layup of the carbon fiber was done The side of the carbon fiber that had been facing
by HPC in Fredrikstad. Due to the asymmetric up from the mold during casting had an uneven
shape of the rear suspension system it was only surface; at most there was a variance of ±1.0 mm
necessary to produce one mold. in the thickness. The design of the knuckle re-
quired both sides to be perfectly even in order to
Three blocks of massive carbon fiber were or- achieve maximum stability between the axle and
dered from HPC. The milling of the knuckles was the knuckle. A 2.5 mm thick layer of Araldite 2031
done at the IPM prototype lab. A STEP - model glue was applied on this side and after hardening
of the knuckle was imported into NX IDEAS 5.0, it was face milled down to create an even surface.
which generated the code needed to mill the
knuckle in the CNC milling machine. A diamond The engine knuckle was milled with a larger
mill with high rotation speed was used to pre- thorough hole for the axle due to the hexagonal
vent delamination and splintering of the carbon shape of the engine axle. An additional insert
fiber. One knuckle was kept as a spare. was required to make it compatible with the
engine axle. The insert was produced by Nomek
The connection points between the knuckle and in Alumec 89 which is a type of aluminum that
the rest of suspension system were constructed has excellent machinability, high strength, low
of threaded metal bars screwed into threaded weight and good stability during machining
holes in the carbon fiber. Araldite 2031 and alu- (Korea Nonferrous metals Co, 2005). It was glued
minum clamps were added to ensure sufficient to the knuckle with Araldite 2031.
strength. For more detailed information, see
Appendix F.

Milling of the knuckle from a massive carbon fiber plate
Glue applied on the knuckles uneven surfaces

metal sheet formed to a tube to prevent glue from entering the axle hole
Face milling of the glue applied to the knuckles surface

3.3.2.3 Lower A-arm connector
L2
Figure 3.3.57: Lower A-arm connector

A
The lower A-arm connectors main purpose is

to transfer the vertical forces that are acting on L3
the wheel over to the coilover. It also keeps the F7
knuckle steady in the lateral and longitudinal F5
direction. This makes the connector one of the
most important parts in the suspension systems,
and also one of the most highly stressed. The L1 F6
connector acts as an extension to the rods that

are going from the knuckle and into the wheel
well. The angle between the rods is what deter-
mines the basic shape of the connector. F1
Weight Optimization and Strength Analysis of the

F1 F1
Lower A-arm Connector
The connector was made in Alumec 89. A lighter Figure 3.3.58: Forces acting on the suspension
solution would be to make it in massive carbon
fiber, but the complex geometric properties between the tire and the ground:
would not make this a suitable choice. The ten-
sion normal to the direction of the fibers could
lead to delamination. A thorough FE-analysis
was performed based on the calculation of the
forces acting on the connector.
Forces acting on the connection joint between It is assumed that the lateral force from corner-
the knuckle and the lower A-arm generated by ing, F1, is equally distributed over the upper and
the vertical force F4 acting on the contact path lower connection joints on the knuckle, hence:
between the tire and the ground:
This yield:
Forces acting on the connection joint between

the knuckle and the lower A-arm generated by
the lateral force F1 acting on the contact patch

F5Y
F5X
F8
F5 α
F9
F6Y
F6
F6X
Figure 3.3.59: Forces acting on the connector
The forces are decomposed with respect to the

B
connectors coordinate system (figure 3.4.59):
C
If added up, this will yield:
A
When substituted, this yield:
F8
F10 z
F9
y x
It is assumed that the longitudinal force, F2,

generated by braking is equally distributed Figure 3.3.60: FE analysis setup for the connector
between the upper- and the lower A-arms,
which gives: A B C
DOF 1 0 1 1
DOF 2 0 1 1
DOF 3 1 1 1
DOF 4 0 0 0
DOF 5 0 0 0
Several FE-analyses were carried out, and the
DOF 6 0 0 0
design was optimized with respect to stress and
deflection. Table 3.3.5: FE analysis set up for the knuckle

Upper rod end
Upper spring
holder Steel washer
Spring
Figure 3.3.61: Test result from the final analysis of

the knuckle Upper POM ring
Containing ring
Value Unit Lower spring holder
F8 1900 [N]
F9 2700 [N] Piston Upper elastomer
stopper
F10 275 [N]
σmax 467 [MPa] Lower POM rings
Table 3.3.6: FE analysis set up and result for the Lower elastomer
connector stopper
Cylinder
3.3.2.4 Coilover
Coilover is short for “coil over shock”, which Lower rod end
means that the spring and the shock is com-
bined in a single, compact package. The main
purpose of the coilover is to absorb the vertical
forces acting on the contact patch between the
tire and the ground, ensuring a smooth ride Figure 3.3.62: Coilover cross section overview
and good handling abilities. The design of the
coilover started very late in the autumn 2011, There are mainly three different types of
and a lot of changes were done during the coilovers, these are the “remote reservoir”, “pig-
spring 2012. See figure 3.3.62. gyback” and “emulsion”. The most common type
is the remote reservoir coilover shock, which is
Theory a mono-tube shock charged with nitrogen. The
A coilover shock is a mono-tube shock with high use of a remote reservoir to house the nitro-
quality that includes provisions to mount coil gen and the floating piston allows the use of a
springs on the shock, which offers a number of shorter tube than would otherwise be necessary.
advantages: The pressure tube is connected to the reservoir
with a short flexible hydraulic hose.
• Completely rebuildable.
• Good tune ability. The coilover allows the A piggyback coilover shock is very similar to the
designer to choose between a vast array of remote reservoir, but instead of the reservoir
spring lengths and spring rates. This makes it being connected to the shock with a flexible
possible to select the perfect spring rate for hose, it is mounted directly on the shock with a
a desired suspension frequency. bracket that incorporates the necessary hydrau-
• The built in spring seat gives the designer lic passage between the cylinder and the reser-
the opportunity to adjust the suspension voir.
height, ride height and preload. It gives
ability to vary between springs of different In an emulsion coilover shock there is no reser-
lengths, hence obtaining different amount voir and no floating piston. The nitrogen charge
of spring travel. is contained in the tube along with the oil in an
emulsion. It is best suited to lightweight and/
or low-speed use, and is also the most compact

type of coilover. Calculation of Spring Force and Spring Travel
The spring travel and spring force are both
The most important part in the coilover is the dependent on wheel travel, wheel force and the
spring, which is designed to carry a load by com- geometry of the suspension system.
pressing. An easy way of describing the physi-
cal properties of a spring is by its spring rate, a It is further assumed that there will be no
description of how stiff it is, expressed in N/mm. change in camber angle during the wheels
The spring rate is defined by four properties: vertical travel. The spring travel as a function of
the wheel travel can then be determined. To
• The spring material simplify the calculations the distance L3 has been
• The diameter of the wire from which the set equaled to zero.
spring is wound
• The coil mean diameter ( the diameter into
which the coils are wound)
• The number of active coils
The effect of the different parameters is listed in

table 3.3.7. If substituted, this yields:
Parameter Change Effect on rate
Wire diameter Increase Increases
Decrease Decreases
Coil mean diam- Increase Decreases
eter Decrease Increases
The spring force as a function of the force acting
Number of active Increase Decreases on the wheel is calculated based on the calcula-
coils Decrease Increases tions from section 3.3.2.3. The value of the force
Fz equals the value of the force F8 determined in
Table 3.3.7: Changes in spring rate
the previous mentioned chapter.
(Ansell, 2008)
Δz3
Δz2 β
α
L3 L2
Δz
Δz1 L1
FZ3
A
FZ2
FZ
Figure 3.3.63: Determining spring travel and spring force

Where:
The distance L3 has been set equaled to zero to • NF = Natural frequency in cycles per minute
simplify the calculations. (divided by 60=Hz)
• SD = Static deflection in inches
(Gillespie, 1992)
The chosen spring gave a frequency of 3.53 Hz,

which is quite high and results in a harsh ride,
but minimizes body roll during cornering. In
comparison a high performance sport car has a
I substituted, this yields: natural frequency of 2 - 2.5 Hz.
Strength Analysis
The most stressed components in the coilover
are the upper and lower spring holder. They are
A wheel travel of ±20 mm implied a spring travel transferring all the vertical force from the wheel
of ±9 mm and the weight of the driver and car to the spring. The design of the coilover requires
would result in a spring force of 390 N. This lead the piston to be able to withstand the verti-
to a required spring rate of 40 N/mm for the cal forces from the wheel if the coilover is fully
spring to be adequate compressed. The spring compressed.
also had to be able to withstand the vertical
force generated by the wheel driving over a A 2.0 mm thick steel washer was added to the
bump (Endresen, et al., 2011) without hitting the upper spring holders design as reinforcement.
top of the wheel well. This implied a necessary A
spring rate of 120 N/mm. A compromise was
made to fulfill these properties. Elastomer
dampers were added in the bottom of the
coilover to help the spring absorb the shock if
the compression force became too high, and a
spring rate slightly higher than 40 N/mm was
chosen.
Property Value Unit
Wire diameter 5 [mm]
F1
Inner diameter 20 [mm]
Unloaded length 70 [mm]
Number of coils 9.8 [-]
Figure 3.3.64: FE analysis of the upper spring
Permitted loaded length for holder.
52.8 [mm]
dynamic load
A Value Unit
Rate 47 [N/mm]
DOF 1 1 F1 1200 [N]
DOF 2 1 δmax 9.3E-04 [mm]
Table 3.3.8: Properties of the SF-TF 1901 compres-
DOF 3 1
sion spring ordered from Lesjöfors (Lesjöfors, 2012) DOF 4 1
σmax 17 [MPa]
DOF 5 1
The static deflection rate of the suspension DOF 6 1
determines its natural frequency. This is the rate
at which the suspension compresses in response Table 3.3.9: FE analysis of the upper spring holder
to weight. The natural frequency can be deter-
mined by a simple formula: The force from the spring on the lower spring
holder will be absorbed by the M18 threads con-
necting it to the cylinder. This was simulated as a
plain inner cylinder wall to simplify the analysis.

Production
A Most of the parts in the coilover were produced
in Alumec 89. The cylinders and the pistons were
ordered from Nomek Maskinverksted due to
time issues. The Upper and lower spring stop-
pers were produced by team members in the
workshop at IPM. Springs were ordered from
Lesjøfors, a Nordic spring company, and the
Spirolox containing rings were ordered from
F1
Smalley, located in Chicago.
3.3.2.5 Toe Link Extension

Figure 3.3.65: FE analysis of the lower spring holder
A Value Unit
DOF 1 1 F1 1200 [N]
DOF 2 1 δmax 3.2E-03 [mm]
DOF 3 1
σmax 29 [MPa]
DOF 4 1
DOF 5 1
DOF 6 1
Table 2.3.10: FE analysis of the lower spring holder.
The analysis of the piston was more complicated

due to the pistons sliding properties inside the
cylinder. There are three sliding connection areas
between the piston and the cylinder when the Figure 3.3.67: Toe link extension
coilover is operating. Only two of these were
applied to simulate the worst case scenario. The toe link extension is used to extend the
A distance between the lower A-arm connection
point and the toe link connection point at the
knuckle, hence obtaining better stability. The
geometry of the knuckle is partly determined by
the mounting points for the brake caliper. These
z
are forcing the connection point for the toe
y
link to be closer to the connection point for the

x
lower A-arm than what is desirable. Any axial dis-
B placement of the toe link rod has a large impact
F1 on the toe angle. This is mainly a problem due
to the 5 mm thick polyurethane (PUR) vibration
Figure 3.3.66: FE analysis of the piston dampers on the mounting points. It is assumed
that the vibration dampers might compress or
A B Value Unit decompress with ±2.0 mm. This gives a change
DOF 1 1 1 F1 1200 [N] in the toe angle of ±2.3° without the toe link
DOF 2 1 1 δmax 4.5E-02 [mm] extension, compared to a change of ±1.1° with
DOF 3 1 0 the toe link extension installed. This is a total
σmax 96 [MPa]
DOF 4 1 0 reduction of 52%.
DOF 5 1 0
DOF 6 1 0
Table 3.3.11: FE analysis of the piston

Coilover assembled
3.3.2.6 Rods and Rod Ends tively.
For more detailed information about the rods

and rod ends, see section 3.3.5.
3.3.2.7 Assembly
Holes were drilled for the mounting points in
the rear wheel well wall. A printed template was
used to find the exact placement of the holes.
A special designed hole-tool was used to make
sure the holes were drilled with the correct angle
relative to the wheel well wall.
The rear suspension system is anti-symmetric. It

means that the holes drilled for the rear suspen-
sion system in the cars body should also be so,
but this was not the case. Due to uncertainties
concerning whether the new motor would be
finished in time or not, the old motor was used
as a backup. It is a lot thicker than the new one,
hence increasing the lateral distance between
the knuckle and the wheel. The wheel would
Figure 2.4.68: Rods and rod ends stick out from the wheel well if the old engine
was used, violating the SEM rules. The design of
Aluminum rods and rod ends were used to the lower A-arm made it impossible to adjust
connect the suspension system to the wheel the lateral distance between the lower A-arm
well wall. The rods were manufactured from connector and the wheel wells wall without de-
aluminum bars (Aluminum 6082) at the IPM creasing the longitudinal distance between the
workshop. They were designed with left handed connection points in the wheel well. The longi-
threads in one end and right handed threads in tudinal distance between the connection points
the other end. The total length could be ad- for the lower A-arm on the engine side was
justed by twisting the rods. The rod ends were therefore decreased, and the holes were drilled
ordered from SKF and the type was SIL 8 C and slightly different than on the other side. The sus-
SI 8 C, left handed and right handed, respec- pension system on the engine side was moved

Paper template with the holes correct position in the rear wheel well
One of the lower A-arm connectors milled out in the Makino CNC milling machine
further into the wheel well. One small block of in the wheel wells. Following the lower A-arm,
aluminum was inserted behind each of the two upper A-arm, coilover and the knuckle was
suspension clevises in the mounting points for mounted, respectively. After assembly the
the lower A-arm to adjust the suspension system coilover needed to be adjusted. This meant ad-
to fit the new motor. The engine was mounted justing the upper and lower stoppers inside the
on the left side of the body since most of the coilover so that the wheels vertical travel was
turns on the race track were left turns. Added restricted to ±20 mm. A wheel made out of card-
weight on the left side would help preventing board was used to do the final adjustments of
the car from rolling over. the rear suspension before the axle, hub, brake
discs and brake calipers were mounted.
The clevises with the belonging wedges and
polyurethane vibration dampers were mounted

Clevises, lower A-arm, toe link, to link extensor and coilover mounted in the right wheel well
From the left: Knuckle with hub assembly, lower A-arm assembly, coilover assembly
3.3.2.8 Further Recommendations Next year’s team should look into the possibility
The rear suspension system worked perfect dur- of making the suspension system as a flexible
ing testing and the race in Rotterdam. However, mechanism, as proposed in the Project Report
there is some play causing small misalignments from the autumn of 2011 (Endresen, et al., 2011).
in the toe angle while driving. Due to time issues The coilover in the existing suspension system
the problem has not been further investigated, is heavy due to the weight of the steel spring. A
but it is believed the cause of the problem is play flexible solution would decrease the weight and
in the Ø10 mm rod end connecting the knuckle reduce the number of moving parts, hence de-
to the lower A-arm. If the suspension system is to crease the mechanical complexity of the suspen-
be used next year, the team should find a solu- sion system.
tion to this problem

Rear suspension with hub assembly and brake caliper
3.4
STEERING
Rotation stopper
Top pulley
Steering colum
Drag link
PU bearing
Bottom pulleys
3.4 STEERING
This is a continuation of the work done on the The chosen concept was a cable based system.
steering in the autumn of 2011. A requirement When the steering column is rotated a cable
had to be added to the requirements list from that is spooled around it pulls the drag link. The
last semester. It was important to have the most chosen cable was made from Kevlar, as a backup
amount of adjustability available for the front solution this cable could be exchanged with one
suspension and steering. This made it necessary made from steel.
to vary the length of the drag link in addition to
having adjustable tie rods (Endresen, et al., 2011).

Thread house
mn
Tie rod clevis
Steering wheel socket
Length adjustment
Cable mount
Figure 3.4.1: All steering components in the final assembly.
3.4.1 DRAG LINK 3.4.1.1 Length Adjustment

There were two fundamental ways of achieving
a variable length: Varying the length of the drag
The overall specifications for the drag link were
link’s main body, or adjustment on each side
done early. They were needed for setting up the
for the tie rod mounts. Lenght adjustments of
steering geometry in correlation with the front
suspension parts are often done by use of double
suspension. It was decided to mainly use carbon
threaded screws with locking nuts. This enables
fiber in the drag link. A pair of LEKI Crosslite RS
fine adjustments to be made.
skiing rods were purchased, dismantled and
stripped of paint. Their properties were well suit-
ed for this application. The decision was made to
use a section of one of these as the main body for
the drag link. Solid plugs are used to internally
support areas where external force is applied.
Figure 3.4.2: Length adjustment concept

Another possibility is to have a sliding mecha- suspension. Rod ends, spherical ball bearings or
nism, either be fastened by friction or secured ball joints are components that can achieve this.
with a locking pin. It is hard to achieve both These parts usually have an angular displace-
accuracy and adjustability with such a setup. In a ment range of ±10-15 degrees. For this solution
continuous friction based binding it is difficult to the tie rod mount had to be designed with this in
maintain a reference point when adjusting. With mind (SKF product catalog, 2012).
a locking pin it is necessary to have pre drilled
holes with certain spacing. The solution from the old DNVFF consisted of a
5mm plain spherical bearing with a machined tie
Threaded bolts are a good solution, but they do rod running through it. To secure the tie rod a nut
have some weaknesses. Screws should not be was used on the other side of the bearing. The
subjected to shear stress. The drag link mainly bearing was mounted in a socket that was fas-
experiences axial load, but some shear force tened to the drag link with an M5 bolt. A casing
can occur at the ends. Load acting on the drag was put onto the socket to secure the bearing in
link will not always be horizontal, but vary with place. A similar solution was modeled in NX.
the compression of the front suspension. Joints
normally used in suspension applications have
a limited angular displacement span. Due to
these problems it was ultimately decided to
have length adjustment of the drag link itself.
The drag link was split in half and two aluminum
plugs were produced. Each plug were manufac-
tured with internal M6 threads, one of them left
handed. These were glued in place in the drag
link with a high strength adhesive. Along with a
6 mm double threaded aluminum bolt, the plugs
made length adjustment possible.
Figure 3.4.4: Old tie rod solution
3.4.1.2 Tie Rod Mount The tie rod diameter was increased from 5mm to
The tie rods connect and transmit the movement 6 mm in this recreation. The specifications for the
of the drag link to the front suspension knuckle. 6 mm spherical plain bearing stated a maximum
Rotation about the x- and y axis needs to be al- angular displacement of 13 degrees. This would
lowed by the interface between the drag link and not be enough.
the tie rod.
As a alternative solution tie rod and mount could
The tie rod has to be able to move with the be split up into two simpler units. Four possible
setups were created and compared to each other.
A simple solution involving gluing a tie rod clevis
in the drag link, as seen in figure 3.4.5.
Figure 3.4.5: Tie rod solution chosen
Figure 3.4.3: Drag link length adjustment

3.4.1.3 Cable Mount 3.4.2 STEERING COLUMN
The steering solution needed two mounting
points on the drag link for the cable. This enabled
The steering column transfers driver input into
the drag link to be pulled from either side. The
steering motion. It is connected to the dash
old cable mounts could be used, as the diameter
board with additional support. In addition to the
of the new drag link was identical to the old.
dash board the steering column had the fol-
lowing interfaces: The steering wheel had to be
mounted to the steering column, transferring
torque. Electrical wires running to the steering
wheel should be concealed for safety and vanity
reasons. The Kevlar cable had to be attached to
the column and spooled around it. The length
spooled per degree of rotation determined the
steering ratio, which had to be within a certain
interval.
The main body of the steering column was from

a 40 mm bolt of aluminum 6061. It was made
Figure 3.4.6: PU bearing, provides low friction for with two different sections. The steering wheel
the drag link section was made with a wide cross section of
26x30 mm to house the wires going to the steer-
ing wheel. The wires should not be seen from the
3.4.1.4 Additional Drag Link Parts
outside of the car or when sitting in the normal
When working on the front suspension geometry
driving position. With the wires running through
an issue regarding the drag link was identified.
the steering column there was no danger for
A critical angle could form between the steering
something ripping them off. The other section of
knuckle toe and tie rod. This would happen if the
the steering column was made with a different
drag link moved beyond a certain limit. To stop
cross section. A certain diameter was required
this from happening two physical stoppers were
to achieve the correct steering ratio. The driver
made for the drag link. These were set diameter
should not have to change grip of the steering
clamps made from Alumec and were placed on
wheel when driving the car. This was achieved
the drag link to limit its axial movement.
by limiting the angle required for full steering
Bearings were produced to support the drag link
lock to ±120°. The drag link had to move about
going through the wheel well wall. Various solu-
±25 mm for the wheels to move from the straight
tions were considered and a simple plastic glide
ahead position to full steering lock. This meant
bearing was chosen. The bearings were produced
that the following diameter d was needed:
from a 40 mm poly urethane (PU) bolt. The flange
in connection with the wheel well wall was tilted
10 degrees to fit perfectly.
Figure 3.4.7: 3D model of the steering column, with sub parts assembled
2 237
The solution was to make the second section The dashboard and steering wheel mount con-
of the column quite small and slide a “thread sisted of two cross beams mounted between the
house” onto it. This unit would be of the needed wheel wells. Plastic cylinders were inserted in the
diameter and have two holes in which the cable core material of the cross beams during produc-
could be threaded into. The housing had an open tion. Holes could easily be cut out afterwards for
end that allowed access to the cable inside. The the steering column. Another set of plastic bush-
cable could be tied around the steering column ings were inserted to form plastic glide bearings.
and the housing could be slid over the knots. The steering column was supported by these two
The housing was fastened with two M5 screws to bearings. Two aluminum stoppers were made for
the column. The cable would be in two lengths, the column. These would press against the bear-
one going to each side of the drag link. If one of ings and stop the column from moving in or out.
the cables should snap it would still be possible One of the stoppers also limited the rotation to
to steer to one side and maintain some level of the 120° specified.
control.
3.4.2.1 Analysis
The maximum force exerted on the drag link by
the front suspension would be 150 N. Friction
in the steering system could double this force.
The pre tensioning of the cable did not affect the
steering column as the forces were in equilib-
rium.
The column was allowed to rotate and move

to a certain extent in the bearing points. This is
because the cross beams were not completely
stiff and some elastic deformations were present.
A load of 300 N was set in the mounting point for
the thread house.
Figure 3.4.8: Thread house The results showed that the steering column
would have no problems handling the 300N load.
For the steering wheel mount it was important The maximum stress was 69 MPa and very small
to make a precise connection, to keep overall deformations.
play in the steering low. This was done by mak-
ing an aluminum socket that would fit perfectly 3.4.3 PULLEYS AND KEVLAR CABLE
into the steering column. The steering wheel was
attached to this socket. A reinforcement ring was
made to support the four attachment bolts on One of the goals for the new steering was to min-
the inside of the steering wheel. The socket was imize friction to improve the feel of the steering,
secured to the column with an 8mm aluminum and enable self-centering. Pulleys would be used
locking pin. together with the Kevlar thread to achieve this.
In 2008 and 2009 the DNVFF used plastic pulleys

with built in ball bearings in its steering system
(Bjugstad, 2008). These pulleys were found in
storage along with some other smaller plastic
pulleys without bearings. The small pulleys were
unused and weighed only 3 grams each. The
larger, heavier ones had a weight of 21 grams
and were in used condition. Both of them had an
internal diameter of 8 mm.
The small pulleys were tested for friction on a

machined aluminum bolt. With some lubrication,
low friciton was achieved, even for relatively high
loads. There was no need to use the large pulleys
Figure 3.4.9: Steering wheel mounting socket with ball bearings.

To mount the pulleys to the body of the car
brackets were produced. Aluminum plates with
M5 bolts going through them were glued directly
to the body as mounting points. Two different
types of brackets were made for the top and bot-
tom pulley mount point respectively.
Figure 3.4.11: Pulley assembled on bracket
Figure 3.4.10: Pulley assembled on bracket
Figure 3.4.11: Strength analysis of the steering column, maximum stress of 69 MPa.
Physical stopper fo limit steering colum rotation to 120 degress.

Kevlar has 5 times the tensile strength-to-weight up than planned since the floor in the car was
ratio than steel. Because of its low bending resist- reinforced with 10 mm core material during pro-
ance less tension was needed to achieve tight duction. This had not been accounted for in the
and precise steering. It could be used with the design process. The consequence of this would
smaller pulleys without any negative impacts be slight bump steer.
such as increased resistance or wear. Two Kevlar
cables were purchased. One small rated at 200 The physical stoppers and cable mounts were
lbs and a heavy duty spear gun rope rated at 400 slid onto the drag link, the cable mounts had
lbs. The heavy duty rope was covered in wax to to be mounted such that their angle was in line
keep it from degrading. with the lower pulley. The tie rod fork had to be
pointing upwards when fastening the mounts.
Tests were conducted to discover the true The entire drag link would rotate when the Kevlar
strength of the cable. A weight scale was at- cable pulled on it if this was not done. This would
tached to one of the roof cranes in the IPM work- compromise the tie rod forks function in the
shop. Weight was gradually added until the cable wheel well. Grease was applied to the drag link to
failed and the maximum load was recorded. Gen- minimize friction.
erally the cable failed in the knots, and at about When the first crossbar of the dashboard was in
50-60% of its rated weight holding capabilities. place the steering column could be inserted. The
This was around 50-60 kg for the small cable and second crossbar was inserted onto the column
over 100 kg for the heavy duty cable. A single and glued in place. Its final position turned out
strand of the heavy duty rope would be used in to be too low. This was because the dashboard
the steering. was more flexible than previously assumed. This
The cable was tested for how much pre tension it didn’t affect the steering columns function, but it
would need to yield accurate steering. A length did add friction to the columns rotational move-
of rope was cut to the same length as the steer- ment.
ing strands. At about 20 kg of load the cable Markings were made for the pulley mount points,
was considered to be adequately tensioned. The using a dummy cable. The bottom pulley had to
knots used to secure the cable would begin to keep the cable lined up with the drag link. The
slip if the tension got too high, this was due to pulley brackets were placed onto the mounting
the wax coating. bolts and secured. The Kevlar cable was fastened
using #8 knots at the cable mounts and bow-
3.4.4 ASSEMBLY AND TESTING ties around the steering column. These were
the two strongest knots found during research.
Holes were cut in the monocoque for the PU The length adjustment of the drag link was used
bearings. The holes had to be made 8 mm further as the base tensioning for the cable. The cable
mounts were used for fine tuning and adjusting

the center position of the steering. 3.4.5 RACE AND CONCLUSION
During the first days of testing it became clear The steering met the SEM requirements by hav-
that the level of friction in the steering was too ing a turning radius of 5 m and very little play in
high. This was caused by the drag link bending the steering wheel. The driver had no problems
from the forces from the Kevlar cable. There was controlling the car during racing. Tension in the
a slight angle between the cable and drag link in cable was kept at an acceptable level throughout
the x-y plane. This caused a small force vector to the 40 minute runs. Between driving sessions the
act normally on the drag link. The length adjust- cable was checked for tension and wear.
ment made the drag link more bendable than it
otherwise would have been. To solve this a third On the first successful test run of the competition
pulley pair was introduced to the car. These were the left tie rod was damaged. This was due to the
mounted on the floor near the bottom pulley. assembly problem with the drag link. The rod
It made sure the cable was parallel to the dag end at the steering knuckle end exceeded its an-
link in all planes. After this addition the steering gular displacement span. This caused the tie rod
performed well. Tension decreased after some to bend. This was an unforeseen consequence of
time due to creep in the #8 knots. By checking the increased height of the drag link. The tie rod
the tension and adjusting regularly the steering was modified to stop this from happening again.
was kept fully functional.
Left: Drag link with length adjustment, cable mounts and physical stoppers. Kevlar thread in yellow.
Right: Upper pulley assembly glued to the wheel well. Visible section of the steering column.
On the final attempt of the competition the
steering failed. The car crashed into the barrier at
the start/stop lane. This was right before the car
should stop for the lap, so the speed was low. The
car was returned to the paddock and inspected.
It was discovered that the left cable had snapped
at the exit of the thread house. The driver had
complained during the run that the steering felt
“hard”. It was harder than usual to turn the steer-
ing wheel. Several things had been overlooked
in the inter race checks that could have caused
this. The track at Ahoy was very dusty. Filth had
collected in the grease applied to the drag link
during the competition. It made the drag link
harder to move. The rear brake hose and electri-
cal wires ran under the drag link on the right side
of the car. They had changed position causing
the physical stopper to catch hold of them.
Straightening the car after a left turn would have
been harder than usual. It is unknown how much
strain this caused on the cable. The mechanism
used to limit the steering columns rotation had
come undone during the attempt. This meant
that the steering column could be rotated further
even though the drag link could not be moved.
The last corner did not require full steering lock
however, so this was probably not a contribut-
ing cause. Finally, the modification done to the
tie rod had one serious side effect. When the left
suspension was compressed the wheel would
get a toe in angle. This would have made the car
harder to steer. All these factors lead to the steer-
ing failure.
The overall result of the steering was good, and

most importantly it satisfied the SEM rules. Its
total weight was less than 1 kg. The complexity
was kept low and it was easy to maintain. Only
1500 NOK were spent in total for the steering.
There were a few goals not met in the require-
ments. The friction level in the steering was
overall too high. The steering column had friction
due to misalignment of the cross bars. The drag
link bent slightly under load even with the third
pulley pair. This caused added friction in the PU
bearings. The high friction meant that no self
centering of the steering was achieved. Despite
the high friction level the driver had no problems
controlling the car. The life span of the steering
was lower than designed. It did last more than six
hours, the time needed to complete the com-
petition. However, this time was started during
the test period and not at the start of the race.
At the end of a run the tension in the cable was
reduced. The requirement for less than 3° play in
the steering was not met during the entire run.

There was at times severe traffic conditions around the track. The driver maintained full control and regularly made
spectacular overtaking moves.
3.5
BRAKES
Master cylinders
Piston
Pedal box
Mounting plate
3.5 BRAKES
The work presented in this chapter is a continu- car and driver. Testing the driver’s maximum
ation of the work done in the autumn of 2011 strength would not necessarily reflect real force
(Endresen et al, 2011). The main challenge for the he or she would exert under emergency brak-
brake system has been to design and produce a ing. A brake failure in an emergency situation
new pedal box and all its sub parts. The rest of could be very serious for the driver, car and other
the brake system is composed of standard, of the contenders. The brake system was therefore
shelves parts (brakelines, calipers and discs). designed with a safety factor of 2 for worst case
scenario.
The brake system is crucial for the safety of both

n arm
Pedal arm
Pistion arm conncetion
Pivot point
Figure 3.5.1: Exploded view of the brake pedal.
3.5.1 BRAKE PEDAL the pedal arm and pivot point.
The new concept implements more support for

The concept developed in the autumn semester
the rods holding the master cylinders. The num-
could potentially have been used in the car, but
ber and size of parts had to be kept at a mini-
there were some concerns. The two rods holding
mum to ensure a light weight construction. All
the two master cylinders were quite long and
the mounting points were integrated in the base
weak. They could potentially bend under load
plate. Small, identical brackets replaced the side
and deform the side walls in the process. Another
walls. This new concept was eventually refined
problem with this concept was that it consisted
into the final design seen in figure 3.5.2.
of too many parts. This would make it compli-
cated to assemble and maintain inside the car,
3.5.1.1 Pedal Arm
the exact opposite of what was intended. It was
Design of the pedal arm directly affects pedal
decided to scrap the entire concept, except for

Figure 3.5.2: Stages of the brake pedal design, from early concept to final design
travel and how hard it is to operate the brakes. Arms of different lengths were connected to the
This component is experiencing the highest piston arm of each cylinder. The cylinder at-
stress in the whole brake system. The master tached to the shortest arm would receive a high-
cylinders had to be pressed with a force of 360 er force due to force equilibrium (see picture).
N, to hold the car stationary on a 20 % slope. The brake balance could easily be fine tuned
For a dynamic case with the car running in wet between the front and rear circuit by altering the
conditions, a higher force is needed to stop the length of the pivot arms.
vehicle safely. With a mechanical advantage of
4/1, 200 N input force from the driver would yield
800 N force to the cylinders. The final mechanical
advantage achieved in the design was 4,4/1.
SEM rules states that the foot plate has to have a

surface area of minimum 25 mm2. The old brake
pedal had a foot plate made of wood with a
surface area of 40 mm2. Producing a foot plate
in carbon fiber would only save a few grams.
The old foot plate was therefore refurbished
and reused. A measurement of the driver’s foot
determined the maximum length of the pedal
arm to be about 220 mm. To ensure that any
Figure: 3.5.3: Pivot point force sketch
driver can operate the brakes, the foot plate can
be adjusted along the pedal arm. The mechani-
cal advantage would however be affected by the
placement of the foot plate.
The car should have more brake force on the For the strenght analysis, NX7.5 was used. The
front wheels to stop the back from sliding, (En- material properties were set to aluminum 6061,
dresen et al, 2011). By having a pivot point in the as Alumec was not available. A force of 600N
middle of the pedal arm, the master cylinders (emergency braking) and constrains was applied
were given different mechanical advantages. in the mounting points.

Figure 3.5.4: The maximum stress of 268 MPa was located around the pivot point hole, which yields a safety factor
of 2. The analysis showed a deflection of 2.5 mm, which will be a conservatiove results due to the material data.
The pedal arm was produced from a 10mm thick be as upright as possible in its resting position.
Alumec plate. Due to problems fastening the This minimizes vertical travel of the pedal, which
work piece in the hydraulic press, the part moved could lead to the driver’s foot slipping off.
slightly during production. This resulted in a
difference in thickness of 0.4mm over the pedal
arm. The performance of the pedal arm will not
be affected by this unevenness. Holes for the
pivot point, mount and foot plate were drilled
out. The hole for the pivot was made to a H7 fit
specification. This way the GE 8 spherical bearing
would not need any external support after being
inserted. 8x10mm aluminum spacers mounted
on a M8 steel bolt were used to adjust the force
arm for each master cylinder. SI 8 E rod ends were
used to connect the cylinders and pivot point
arms. Spacers and rod ends were secured in place
by fitting a nut to the end of the bolt.
3.5.1.2 Brake Pedal Box

The brake pedal box has to withstand the op-
posing forces of the brake pedal arm and master
Figure 3.5.5: Placing of master cylinders and pedal
cylinders. In addition to being strong, it needed
arm. From this the minimum distance between
to be securely fastened to the car. It is the most
components was determined.
complex sub-part of the brake pedal and re-
quired the most attention. The design process continued with removing ma-
terial from a rectangular slab of 145x60x30 mm,
It was decided to go for a simple, light and strong just as one would in a milling machine. Because
design for the pedal box. The pedal box needs of the high forces involved it was desirable to
to accommodate adjustments of the master have the cylinder piston arms as straight as pos-
cylinders, due to variable length of the pivot sible towards the pivot. Some cross bracers were
point. In addition required piston travel had to left between the walls to keep the ‘box’ rigid.
be achieved. The pedal arm needs to stop nor- Edge blends were applied to avoid any stress
mally to the piston arm of the master cylinder concentrations. The bracers made sure that the
when pressed. This will maximize the mechanical part was not damaged when it was fastened in
advantage during braking. The pedal arm should the milling machine’s hydraulic press.

Figure 3.5.6: Grooves removed from block and final piece
Underneath the box 2 mm deep grooves simi- 2011). Alterations had to be made to the master
lar to those on top were subtracted to reduce cylinders for them to be compatible with the rest
weight. Two 6 mm bolts going across all four of the system. An adapter solution was made for
walls would hold the master cylinders in place. the hydraulic hoses going to the calipers. Bicycles
The pedal arm was held in place with an 8 mm use 6 mm hoses and banjo plugs for the brake
steel locking pin. Two mount points were left in system, while motorcycles use 10 mm. The solu-
the floor of the box near the pedal mount. Two tion was to produce a M10 bolt with internal M6
bolts would be used in these points to fix the box threads. This bolt could be secured and sealed
to the car. with a gasket in the master cylinder. The original
6 mm banjo screws could then be used without
The analysis was performed on an earlier revi- modifications. To reduce weight the M10 screw
sion of the part. Another brace would be added was machined out of aluminum, and the piston
to the design before production. The results of arms were cut to half their original length.
the analysis were conservative and still valid.
Aluminum 6061 was assigned as the material. To further reduce weight the cylinders were fas-
Forces acting on the master cylinder does not tened in the milling machine and excess material
act directly on the holes supporting the cylinder was removed. This saved 35 g per cylinder, but
cross bolts. This was simulated by adding 1D con- it was discovered that they had been damaged
nectors, as seen in figure xx. To simulate brake by the hydraulic press. Some deformation and
balance, 1700 N and 900 N force vectors were set cracks had occurred on top of the cylinder at the
on the two master cylinder points. For the pedal connection of the brake fluid reservoir. The cylin-
arm supports the reaction forces were simulated ders were fixed by filling the cracks with Araldite
by bearing forces of 2000 N. The pedal box was and machining the hole circular. After the re-
held in place by a fixed constraint on the contact pairs the cylinders were tested and found to be
area of the two mounting bolts. A simple support in working order. The other area for potential
from below simulated the floor of the car. weight savings was exchanging steel parts with
aluminum. Both the piston arms and hydraulic
The brake pedal box was machined out from a connection bolts were made from steel. Remak-
single block of Alumec, in the Makino milling ing these parts in aluminum saved about 100 g.
machine at IPM. According to volume calcula- A total of 170 g was trimmed from the master
tions done in NX the finished part would weigh cylinders.
approximately 120 g. The Alumec block had a
mass of 1 kg before production started. Holes for 3.5.2 OTHER BRAKE CIRCUIT PARTS
the master cylinders and pedal arm were made
with H7 specifications. The front most bolt would
The rest of the brake system consisted of stand-
carry most of the load from the master cylinders.
ardized, off the shelves parts. Some were pur-
A steel bolt was used for the front mount, while
chased and some were reused from the old car.
an aluminum bolt would suffice for the rear. The
As the products for bicycle brakes already have a
pedal arm locking pin was made from steel in the
low weight, very little can be gained by making
turning machine. The final piece had a weight of
custom parts. The only area where a significant
116 g (without bolts), a very satisfactory result.
weight reduction could have been made was the
steel brake discs. These could have been replaced
3.5.1.3 Master Cylinders
with discs of carbon fiber, but this was not prior-
More detailed descriptions and calculations
itized.
performed on the new master cylinders can
The hoses reused from the old car were of a steel
be found in the autumn report (Endresen et al,

Figure 3.5.7: The maximum stress was 136 MPa, giving a safety factor of 4. This stress occurred at the pedal arm
support. There were no significantly deformations in the pedal box
Figure 3.5.8: Displacements in the pedal box

Left: Master cylinder with piston arm and hose connecting the fluid reservoir
Right: Avid Elixir 5 brake caliper, 160 mm rear post mount adapter and brake disk
reinforced type. This reinforcement stops them spaced out to solve this problem. Because the
expanding under load, giving a firmer feel to discs are attached to the hub they had to be
the brakes. DOT fluid used in the brake circuit spaced an equal amount. Spacing the discs was
required o-rings made from a special rubber. The done by stacking three M5 washers per bolt
reused calipers were from two different brands, (each disc had six bolts). To do this accurately a
Formula Oro and Avid Elexir 5. The Formula micrometer had to be used.
calipers were the lighter of the two, and there-
fore placed in the back. They both had the same Hydraulic hoses were connected from the
piston size of 22 mm. calipers to the master cylinders and the system
was filled with DOT 5.1 brake fluid. Bleeding
A more detailed description about the status of the brakes was a simple process that could be
the parts reused can be found the report from performed by two people. One had to actuate
previous semester (Endresen et al, 2011). the fluid syringe connected to the brake caliper.
The other would tell when the level in the fluid
3.5.3 ASSEMBLY AND TESTING reservoirs started to rise. When this happened
the circuit should be air free and full of fluid. The
bleed kit was purchased by an earlier SEM team.
After the brakes had been assembled a quick
function test was performed. Because the mount-
When working on the suspension the calipers
ing holes on the master cylinders were elongated
were often taken off. To stop any fluid leaking
they could be moved back and forth about 5mm.
they were fitted with bleed stoppers. These stop-
This presented a solution for the brake sensor for
pers compress the caliper piston fully, to stop
the electrical system. A bracket for the sensor was
movement if the brake pedal is pressed. The cali-
made and fastened using a hole made for weight
pers needed to be fine adjusted to stop the discs
savings. This positioned the sensor directly in
from rubbing the brake pads. Using a bright light
front of the cylinder mount. By fitting the pedal
underneath the suspension made fine tuning
arm with a powerful spring it would pull both
easier. It was possible to visually determine which
master cylinders towards it, thus trigging the
way the caliper had to be moved in order for the
sensor.
disc to rotate freely. Two of the discs were slightly
bent making it impossible to eliminate scrubbing
When mounting the calipers and discs a prob-
completely. This was only crucial for the race, so
lem was discovered. The wheel hubs were not
the old discs were used in the testing period.
designed big enough in the axial direction. This
was exaggerated by the fact that the backside of
During the test day at Dragvoll several test were
the rims was rough and uneven, so the wheels
done on the brake system. Two team members
touched the brake calipers. The hubs had to be
were needed to test the car. One person pushed

the car up to a certain speed and the driver so no scrubbing occurred. Some deformation of
applied the brakes. This speed was gradually in- the discs will occur during operation. After the
creased from run to run. The brakes stopped the discs had been used for a run the calipers needed
car every time in a very short distance. As a final to be readjusted. Some spillage was detected
test the brakes were ‘slammed on’, which resulted around the brake pedal. It was discovered that
in the rear tires actually locking up. The brake back pressure will build up in the fluid reser-
balance at this point was 66/33% (front/rear). voirs. The seals around the cylinder pistons are
The conclusion of the test was that the braking not completely tight. Some fluid will escape the
systems performed quite well. After adjusting the pressure circuit every time the brake is pressed.
brake balance forward to stop the rear wheels This leakage is very minute and did not affect the
from locking up, the performance was excel- performance of the brakes. Some problems with
lent. Several more tests were conducted with no the brake sensor were also detected. The spring
further issues. pushing back the pedal arm was worn and not as
stiff as when first installed. To trigger the sensor,
When working on the car in the workshop, two the brakes needed to be released abruptly. This
of the brake discs were damaged. One case was more a source of annoyance than an actual
involved the engine wheel, where the disc is di- problem on track. The driver had no problem
rectly attached to the rim. When taking the wheel operating it.
off the car, the brake caliper first needs to be de-
tached. This was forgotten and one of the good All obligatory requirements were fulfilled. The
discs was bent. Dymo notes with warnings were exchange time for the parts worked on was
put in positions were they could be seen when less than 30min. Total cost for the brake system
working on the suspension. No more incidents was 6000 NOK. The total weight of the system
occurred after this. New brake discs would not be was less than 2kg. A 50% weight reduction was
installed until the car had reached Rotterdam as achieved for the brake pedal compared to previ-
a precaution. The system was otherwise ready for ous year. The new brake pedal had a final weight
competition. of 750g. Total number of parts far exceeds 50,
though the number of major parts is below 50.
3.5.4 RACE AND CONCLUSION
For the race attempts the new brake discs were
mounted. The calipers could now be adjusted
Brake pedal assembly with reservoir holder and brake sensor. Mouting plate glued to the floor of the car.
3.6
WHEELS
Outer CF layer
Massive CF core
Valve
Rim ring
Inner CF layer
Tyre Foam core
Figure 3.6.1: Exploded 3D CAD model of the rims produced in 2012
3.6 WHEELS
The development of the rim design started in the An internal pressure of 5 bars was applied on the
project phase executed in the autumn 2011. It rim rings surface.
was finalized during the spring 2012. A
DOF 1 1
3.6.1 STRENGTH ANALYSIS DOF 2 1
Table 3.6.1: FE analysis set up for
DOF 3 1
DOF 4 1
the rims.
The strength analysis of the carbon fiber layup
DOF 5 1
was performed in UGS NX Nastran. The carbon
DOF 6 1
fiber type used was DB 420.
B
Figure 3.6.2: FE analysis set up of the rims.

Figure 3.6.3: FE analysis results of the rims.
At point B and C symmetry planes were applied.

3.6.2 PRODUCTION verksted at NTNU.
The hole for the valve was drilled on the rim rings
The mold in ebaboard was milled at To-mo
inner surface, which had the most free space due
Modeller og Interiør AS in Fredrikstad. This was a
to the rims asymmetric shape. It was placed there
positive plug which was used to produce a nega-
to prevent the valve from coming in conflict with
tive mold in fiberglass. The result was an accurate
the tire. A special jig was made to secure the rim
and durable negative mold that could be used to
while the hole for the valve was made in the drill
produce a large amount of rims before its condi-
press, and to make sure the hole was drilled with
tion would be exacerbated.
the correct angle relative to the rim rings surface.
The hole was drilled with a Ø4.0 mm bore, fol-
The rims were produced at HPC by using the
lowed by Ø8.0, Ø10.0 and Ø11.5 mm which is the
same layup method as in 2011 (Qviller, Stock-
specified bore size for the valve. This was done
fleth, Bleie, & Hoel, 2011). 8 layers were applied,
to avoid delamination and rough edges in the
giving the rim an overall thickness of 4.0 mm.
carbon fiber.
Three regular rims and one engine rim was made,
plus one spare of each.
Small cracks and irregularities in the rim rings
inner surface were filled with Araldite 2031. After
The center hole and the bolt circle on the regu-
applying the glue and sanding it down with
lar rims were made by HPC, but the 45° chamfer
fine sand paper the rim rings inner surface was
around the center hole had been left out from
smooth and without any visible cracks or irregu-
the production. This was a crucial part of the rims
larities.
design, and without it the rims would not fit onto
the hubs. The chamfer was made by Finmekanisk
Drilling of hole in the rim ring for the TR-414 tubeless snap in valve
Left: Araldite 2031 to cover up small cracks and irregularities on the rim rings inner surface
Right: Sikaflex and PVC tape to prevent leakage
The crack in the broken rim after testing at Dragvoll

Testing showed that the rims were not airtight, free. Testing showed that the rim was able to hold
loosing pressure over time. A closer investigation a pressure of 5 bars after the repair, but it was not
revealed that air was leaking through the carbon used during the competition.
fiber in the rim ring. This could cause the rim ring
to delaminate. Air could also penetrate into the The front left rim exploded on the way back from
rims center plate and cause a delamination be- Rotterdam. The incident happened after the car
tween the carbon fiber and the foam core, causing had been standing outdoors in the sun, and it is
severe damage to the rims. A thin layer of Sikaflex reason to believe that the energy from the sunlight
was added on the rim rings inner surface. This had heated up the air inside the tire. The internal
created an airtight membrane and stopped the air pressure was initially 5 bars and the heated air may
from leaking through the carbon fiber. A protec- have generated a high enough internal pressure to
tive layer of PVC tape was applied to prevent the destroy the rim. Next years team should be aware
Sikaflex from being damaged and further testing of this and not place the car in the sun without first
showed that the rims were performing well and decreasing the tire pressure. The rim has not yet
without any leakage. been repaired.
3.6.3 PERFORMANCE The rims produced in 2011 are each 200 grams
than lighter the ones produced in 2012. Therefore
the rims from 2011 were modified to fit the new
Testing revealed that the rims were not as good as
hubs, and were used on the two last attempts in
first predicted. The unevenness on the rim ring was
Rotterdam. They performed well and could also be
the main problem, causing the wheel to wobble
used next year. However, it is recommended to use
in the lateral direction when rotated. This lead to
them as spare rims only.
reduced handling abilities and loss of energy while
driving.
3.6.4 FURTHER RECOMMENDATIONS
One rime cracked after the first outdoor testing
day at Dragvoll. It happened after the car had been Next years team should make new rims. It is highly
transported back to the workshop, and the tire recommended to produce them in another mate-
had an internal pressure of 5 bars. It was reason to rial than carbon fiber. Several of the teams in Rot-
believe that something had gone wrong during terdam had rims milled out of aluminum. Produc-
the production process of this rim since the rim did ing rims in carbon fiber is a complicated job and
not explode, but air was leaking slowly through several risks will be eliminated by producing them
a visible crack in the rim ring. The damaged rim in a more durable material.
was sent to Paal Fediuk at HPC who repaired it for
3.7
PROPULSION
Figure 3.7.1: Exploded view of comlpeted motor assembly
3.7 PROPULSION
3.7.1 MOTOR DESIGN Design of the stator, magnet arrays and air gaps
are based on project report of Endresen (2011).
This section summarizes the work done on the The stator is a resin cast with Litz wire windings
motor parts. It is an improvement of last year’s cast inside. Torque is trasferred to the rim tho-
motor based on the specifications from last se- rugh the rotors.
mester (Endresen, 2011).
3.7.1.3 Hub and Axle
3.7.1.1 Requirements The new magnet array, with the removal of the
The motor was given a budget of 100 000 NOK. In iron core rings used in previouse years, the motor
addition to service hours, SmartMotor sponsored can be made much more compact.
the purchase of the magnets, which were ex-
pensive and had a lead time of many months. A The diameter of the axle is designed to accomo-
significant improvement over the previouse iron date easy insertion of wires for the three-phase
ring motor is the use of a Hallbach array which drive current. It is, however, kept reasonably thin
eliminates the need for steel cores to conduct to avoid oversized and heavy bearings.
the magnetic flux and reduces the weight of the
engine considerably – if the magnets are glued Larger diameters enables a octagonal section
directly onto the rotors. that mates with the suspension. This will make it
simpler to attach the motor wheel to the car as it
Due to the efficiency of the motor, combined does away with the key that easily fell out
with the low nominal output (around 100 W) al- (Bleie et al., 2011).
leviates the need for cooling, it is therefore not a
requirement.
3.7.1.2 Concept Description

In general the Hallback array motor is an incre-
mental improvement of last year’s motor, and its
concept is therefore largely already determined.

Figure 3.8.2: Cross section of motor
Requirement Value Must Should

Withstand lateral force caused by cornering F1 X
Withstand longitudinal force caused by braking F2 X
Withstand vertical force caused by weight of car plus
driver, and g-forces F4 X
Withstand torque from braking X
Fit the engine rim X

Allow space for the three wires for the drive current X
Adjustable magnet air gap X
Non-magnetic material X
Support the strong magnet array
4000 N X
No parts interfere during rotation X
Separable X
Weight < 10 kg X
Weight < 7 kg X
Table 3.7.1: Requirements specification for mechanical motor parts

Figure 3.7.3: Close up of the motor hub
Figure 3.7.5: Load cases for an axle during

cornering
Figure 3.7.4: 3D view of the new motor axle
The motor axle will be used on the left side of

the car, and as the sharpest turns go to the left,
the motor axle will rarely undergo the extreme
loads of hard cornering. The spreadsheet shown
in appendix E was used to determine the forces
applied to the inner or outer wheel of a typical
curve. This is important because the lateral force
developed in the contact patch is low when it
acts to further bend the axle. When the force is
high due to turning in the other direction, the
force pushes the car up and the bending Figure 3.7.6: FEM analysis of motor axle subject to a
moment decrease (see figure 3.7.5). hard bump

Figure 3.7.7: FEM analysis of motor axle on inner
wheel of a typical, sharp curve
Figure 3.7.8: FEM analysis of motor axle on outer

wheel of a typical, sharp curve
The simulations of the motor axle were

performed without the support of the cone
(figure 3.7.1) as the conditions were difficult
to reproduce in NX. With the cone in place the
stresses of 190 MPa from the bump simulation
will be drastically reduced.
The axle was machined by Finmekanisk for 5 400

NOK, using Aluminium 7075 provided by the
department. Figure 3.7.9: Outer and inner rotor with colorized
magnet arrays
3.7.1.4 Rotors
The rotors consit of circular plates with mag- 3.7.1.5 Rim Adapters
nets glued on. Bearing with bearing housings is
placed on both plates. Due to the axles geom-
etry, a smaller bearing is used on the outer plate The rim adapters allows the motor to be mount-
to reduce frition and weight. ed non-permanent into the rim ring. Design of
the array motor is smaller in diameter than the
Figure 3.8.9 shows the rotor plates. The outer previous iron ring motor. This allows two conical
rotor (top), shows a protection ring which will be rings to be inserted from each side of the rim,
used since the rotor is made from carbon fiber, to and 16 M4 bolts keeps the motor in place.
protect the carbon fiber from the bolts used to
separate the motor. The three holes (bottom
plate) are threaded with M10 to allow the bolts to
be inserted and force the rotors apart. Because of
the strong magnets it is not possible to separate
the motor otherwise. Since the plates are made
in carbon fiber the three holes for separation
would be made by threaded inserts.

Figure 3.7.10: Close up of rim adapters, in green and blue. The yellow section is the rotor spacer
Since the outer rim adapter will be exposed to tip, which can be inserted into an normal power
sunlight, it must be from a UV-resistant polymer, drill.
or else be coated with a UV-blocking paint. Dur-
ing competition the rear wheels will be occluded
by aerodynamic covers anyway, but experience
from testing shows that the car is tested a lot
without wheel covers.
3.7.1.6 Test Rig and SM-Adapter

Testing the motor thoroughly is important, and
SmartMotor’s laboratory offers a test bench and
virtually unlimited service hours. To be able to
test the motor, an adapter must be produced. An
aluminum adapter was produced to fit the exist-
ing test bench configuration at SmartMotors.
The adapter allows the motor to transfer torque

to an opposing motor which can be used to
either drive the motor or measure the energy Figure 3.7.11: The new test adapter
output.
Experience from the SEM team of 2011 revealed

the need to drive the motor at cruise RPM to
measure the back-induced voltage during
competition. The adapter is equipped with a end

Top: New motor axle
Bottom: Assembling the new motor
3.7.2 MOTOR DEVELOPMENT
Both the motors are axial flux, permanent mag-
net motors with no iron in the stators. The prin-
cipal difference is to use a magnet arrangment
called Hallback array instead of iron rings to lead
the magnetic field. A benefit of Hallbach array is Figure 3.8.15:Three phase single layer wave winding
that the flux lines are more concentrated giving a
higher flux density. Also the iron ring motor has
a stator with single layer distributed windings
made of enameled copper wire. The Hallback
array motor has a double layer wave winding
made with litz wire. There is however a great dif-
ference in size and weight. The iron ring motor Figure 3.8.16: Single phase distributed winding
has a weight of 17 kg, compeard to the Hallback
array motor that weighs 7 kg. Large ammount of
iron in the motor is the biggest contributor to the
large difference. The rings location leads to a high
moment of inertia.
Figure 3.7.17: Three phase distributed winding
3.7.2.1 Rotor
The rotor consists of carbon fiber plates and per-
manent magnets. Magnets were glued on with
a two component Araldite. The glue was recom-
mended by one of our major sponsors with a
wide experience within composites. Before the
magnets were glued, the Araldite was tested
for tensile strenght. Four tests were performed.
Three of these showed acceptable results, the last
showed insufficient glue in the necessary area.
Figure 3.7.12: Magnets in a Hallbach array, arrows The successful tests showed that the carbon fiber
showing direction of magnetization delaminated before the glue failed. This makes
the carbon fiber the limiting factor.
The magnets used were NdFeB N42 magnets.
These were selected because of their magnetic
strength and their thermal abilities. The mag-
nets were assembled in a Hallbach array before
they were attached to the rotor plates. Several
attempts were made in order to find the right
technique for doing this. The first attempt was to
use a circular wooden plate with another wood-
en plate of a smaller diameter centered on top of
the first. Shape and diameter of the array can be
Figure 3.7.13: Magnets and iron ring, arrows show-
controlled this way. This procedure was however
ing direction of magnetization
unsuccessful as the magnetic field caused the
magnets to turn out of position. By adding a third
wooden plate on top of the center plate, the
problem was fixed.
The plates were made of wood since wood is

not magnetically conductive. When the magnets
were arranged in the Hallbach array they kept
Figure 3.7.14: Single phase single layer wave wind- themselves in position. The magnet arrangement
ing collapsed several times due to disturbances. An

Assembly of Hallbach array
important challenge was to keep all metals in ment had to be constructed. This was particularly
a safe distance. Since the plates were fastened important when gluing the magnets as the forces
with a bolt and nut, wrenches were necessary. here would be especially high. When the mag-
Careless handling of these caused the magnet nets were assembled in the Hallbach array, glue
arrangement to collapse. was put on the carbon fiber plates. The plates
were then lowered onto the Hallbach arrays. In
The rotor plates were made of carbon fiber. These order to get the magnets centered on the ro-
were milled out to the correct thickness and with tor plates, wide shafts were designed to exactly
the proper flange to support the spacer ring and match the attachment holes in the rotor plates.
magnets. The necessary thickness was found by These shafts were then put on the bolts in the
first calculating the forces that the plates would center of the wooden plates used for the magnet
be exposed to. HPC suggested a thickness of assembly. This ensured a right possition of the
2 mm would be sufficient. To increase safety, a magnets. In addition its kept the carbon fiber
thickness of 4 mm was chosen. plates horizontal while lowering. This led to a suc-
cessful distribution of Araldite 2021 glue.
To assemble the rotors properly, special equip-

Assembly of Hallbach array
Rotor production
Welding of stator mold
3.7.2.2 Stator Production air, epoxy and wires. In the bottom plate, four
The stator consists of litz wire and epoxy. The bolts were attached. These ran through holes in
wire was wound on a wooden board. To ensure the remaining pieces. Bolts were intended for
a clear layout of the wires a technical drawing tightening the plates together during the vac-
was printed and stapled onto the wooden board. uum process. In order to get the cast out of the
Nails were used to keep the wires in place. With mold after casting the mold needed to be cov-
three phases and 24 pole pairs, 288 nails were ered with release agent. This was smeared on the
needed. Another wooden board was used to pieces which were then baked at curing tempera-
make an outer radius guideline. This was neces- ture for 10 minutes, three times. The pieces were
sary to ensure the correct size of the coils. A mar- baked so the release agent would set properly to
gin of 1 mm was added to the diameter. Six wires the steel and to prevent it from reacting with the
were then wound in a double layer wave wind- epoxy.
ing. The winding was then tied up with 288 knots
to replace the nails. Cotton string was used to tie The second mold was of a simpler design, con-
the winding as this is not magnetically conduc- sisting of Ebaboard with a stator sized hole milled
tive. Steel wire was also considered as this would out. This was then covered with spray paint and
have an easier assembly but this was discarded wax. Spray paint was used to fill the pores in the
as it would lead to a higher leakage flux. The nails material and the wax was a release agent suitable
were then removed. for low temperatures. The same lid was used for
When the wires were wound, tied and properly this mold to squeeze the wires.
fitted in the circumference ring, the wires were
placed in the mold. 3.7.2.4 Open Casting
The open casting technique used the ebaboard
3.7.2.3 Mold Production mold. Windings is placed in the mold with wire
For the different stators two different molds were ends coming out of the top lid. Epoxy is poured
made. The first mold was made for vacuum cast- in and the lid is pressed down. This causes some
ing. This was a four piece mold made of steel. The of the epoxy to overflow the mold. Before pro-
mold consisted of a plane bottom plate with an duction, too little epoxy was prepared, resulting
entry hole for the epoxy, two semi-circle shaped in shortage of epoxy. Still, this stator had many of
side plates and a plane top lid with exit hole for the desired properties.
Open casting of the second stator
3.7.2.5 Vacuum Casting with sealing tape, but the mold was particularly
The method of vacuum casting was recommend- challenging to seal. This might have been due
ed by professor Nils Petter Vedvik at IPM, as a to the welding done when making the mold, as
casting method suitable for electrical equipment. this may have created an air canal from the side
This method will, if performed successfully, give of the mold. The distance between the pressure
great dielectric properties and prevent partial bolts was too long for supplying the right pres-
discharges. In this design partial discharges is not sure in between the bolts. This problem was
an issue and it would therefore be acceptable solved using clamps along the circumference of
with minor imperfections in the cast. However, the mold. In order to check the cast procedure
casting the stator was quite challenging. several test stators were made. When the first test
stator was made it included three wires coming
A vacumm pump was connected with tubes to out of the mold for simulating the power cables
two safety chambers in series, before it was con- and a bundle of litz wire was also tied tightly and
nected to the topside of the mold. The mold was put in the mold. This was done in order to exam-
slanted so the outlet tube was position higher ine how the epoxy would surround the litz wire.
than the inlet tube. This configuration enabled
the mold to be filled with epoxy in a controlled 3.7.2.6 Spacer Ring
manner. The epoxy would flow into the mold and To keep the rotor discs at a desired distance, a
it was easy to see when the epoxy would come spacer ring was constructed. The ring was sup-
out of the topside. When this happened the ported by a flange in the rotor discs. In order to
outlet tube was blocked and the epoxy would achieve the right mechanical strength polyoxy-
continue to fill the mold until the pressure inside methylene (POM) was selected as material. The
the mold was equal to the atmospheric pres- flanges in the rotor discs were highly necessary
sure. One major challenge here was to keep all as there was a very small margin for misplace-
interfaces airtight. All connections were covered ment. Any mispalcement of the space ring would

Left: Air bubbles in the first test stator produced with vacuum casting
Righ: Preparing the vacuum casting of the first stator
lead to rubbing between the spacer ring and reduce the wear on the outer plate, an aluminum
stator. ring was glued to the outer plate were the rods
would come in contact. With all this equipment
3.7.2.7 Assembly of the Motor in place the inner plate could be lowered onto
The stator was to be fitted on an aluminum axle the rest of the motor.
made especially for this project. Necessary holes
needed for fitting and securing the stator on the 3.7.2.8 Problems Regarding the Construction
axle were milled out. When the motor was assembled the axle was
spun. Clear cogging force was experienced, simi-
The stator was mounted on the axle and the lar to that one may find in iron based machines.
three power wires were lead through the hollow Since the motor is completely ironless, this indi-
axle. Wires would then come out on the inner cated a short circuit in the stator. The motor was
side of the engine, the side facing the car. then brought to Smart Motor for testing. Using a
Rotor plates were fitted with ball bearings in the separate engine, the motor was spun as a gen-
center holes. The axle was then put through the erator and torque was measued. Even with open
ball bearing in the outer plate and the spacer wires the machine still needed 7 Nm just to spin
ring was then put on, surrounding the stator. at rated speed. Since there was no mechanical is-
Last, the inner rotor plate was to be lowered onto sus regarding the motor, the motor was disasem-
the remaining parts. Since the magnetic forces bled and the stator removed. The motor now
are very strong, this process needed special tools. spun freely. This showed that the problem was
Five holes were made in the inner rotor plate and stator related. The stator was inspected visully
fitted with threaded inserts of aluminum. These and by loading it with a high current while exam-
inserts would allow assembly of the motor plates ining it with a thermal camera. Still, no problem
with threaded rods. The rods went through holes were found.
in the stator and down to the outer rotor plate. To

Thermal camera used to examine the first stator when it was exposed to high current
After this a new stator was constructed. Due to When the measurement was done the windings
the problems with the first stator it was a high were reconnected in the correct order.
priority to keep all connection points out of the
epoxy. One severe problem with this stator was that
since all twelve wire ends needed to come out
Since the outer shielding of the litz wire was of the cast, the wires needed a lot of space in
made of silk and nylon, it was not airtight. It the axial direction. This led to a bigger air gap
would therefore be difficult to use vacuum than what the engine was designed for. Since
casting and get the wire ends out of the mold. the torque is τ=B×I, more current is needed to
Therefore, the production method was changed achieve the same torque. The loss is dependent
to open casting. After the new stator was casted, on the current Ploss=R×I2. A bigger air gap lowers
it was assembled and tested. Tests showed an in- the motors efficency.
duced voltage much lower than anticipated. The
connections were then opened and each wire
end was connected to thinner wires which were Due to limited time, the Hallback arry motor was
lead through the axle. This enabled monitoring of not finalized, there was therefore decided to
the voltage induced in each winding. All the in- continue using the iron ring motor. Further test-
duced voltages were as expected. This concluded ing and optimalization was therefor initiated at
that the wires had originally been connected so SmartMotors.
the induced voltages counteracted each other.
Left: Engine with all wire ends accesable

Right: Induced voltages in the new engines final winding setup
Damaged litz wire
In order to burn off the power supplied by the

engine from the battery, three resistors were con-
nected in a delta-connection. These were then
connected to the terminals of a load engine. This
was done because the electronics normally used
to drive the test engine were not able to receive
the induced power.
Figure 3.7.18: Winding connection
In between the motor and the test motor a
torque meter was connected. The torque was
3.7.2.9 Testing the Existing Motor
controlled by adjusting the resistors. This solution
The tests were performed using the battery
supplied constant torque for all steady states, but
intended on board the car. The air gap of the iron
changed with the speed. This is due to the fact
ring motor was decreased as planned (Endresen,
that the induced voltage is proportional to the
et al., 2011). The project is sponsored by two sup-
speed, while the counter torque is proportional
pliers of different batteries. It turned out that the
to the current. The current is a function of volt-
battery pack with the highest number of Ampere
age and resistance. Therefore it was impossible to
hours had a significant voltage drop. It was cho-
check the transition between operating points.
sen to increase the air gap as a middle ground
and this seemed to work. It was later discovered
that the other battery pack had a much lower
voltage drop. Unfortunately this battery pack had
a lower energy capacity than the first battery. A
third battery pack was therefore ordered which
had twice the capacity of the second.

Resistor setup
Figure 3.7.19: Efficiency as a function of torque of the old engine at 300 rpm, 31,55
km/h as it was in2011
Figure 3.7.20: Efficiency as a function of torque of the new engine at 257,5 rpm, 26
km/h with the smallest obtainable air gap
The efficiency of the drive train is limited by had an air gap adjusted for the first battery pack
two factors, air gap and back induced voltage. with a high voltage drop, the air gap was reduced
Lowering the maximum speed will increase total to better match the second battery pack with
efficency. a more stabile voltage. This was done by first
reducing the air gap, then using a drill to spin
3.7.2.10 Conclusion the engine at nominal speed, measure the back
The car had not been tested under windy condi- induced voltage and implement the new param-
tions. When the car was accelerating facing the eters in the engine controller. No more changes
wind, this required more current than the battery were made to the engine for the rest of the
was able to deliver. The battery management competition.
system shut down and the car stopped. This was
solved by rewriting the code in the acceleration The Hallback array motor is operational but it is
program to have a lower acceleration. not as energy efficient as it should. This could be
greatly improved by making a new stator. Experi-
The engine was also modified. Since the engine ence shows that the stator width could probably

be reduced from 8.7 mm to about 6 mm. The makes it easier to perform tests on the windings.
wires should not be taken out of the stator cast. One test that is interesting is the Megger test.
Instead the wires should be soldered onto con- Here a high voltage is applyed across two wind-
nection rings that are casted into the mold. When ings that are not connected and measure if there
the cast has cured, one can drill away the epoxy is any current flowing. This is interesting when
to free these rings and thereby connect the working with litz wire since partial short circut
windings without using too much space in the may occure.
axial direction. It is also recommended to keep all
twelve wire ends available for connections as One severe issue with the existing drive train is
this makes it easier to make changes if mistakes the moment of inertia of the motor. The old mo-
are made. Another advantage of this is that the tor consists of big magnets and a heavy iron ring
windings can be connected in parallel instead of placed for away from the axle. This makes is easy
series as they are now. This would allow a higher for the engine to create a high torque but it also
current and thereby a higher torque. This could gives a high moment of inertia. The new engine
be useful during acceleration. has a lower total mass with smaller magnets and
no iron. This gives a lower moment of inertia but
it should be considered if a new engine setup
could be preferable. The problem with having a
high moment of inertia is that it requires a lot of
energy during acceleration. In 2012 the compe-
tition had a shorter track than before and the
driver had to do more laps with a complete stop
in between in each lap. This made it more neces-
sary than ever to reduce the needed energy for
the acceleration process. Also the engine control-
Figure 3.7.21: Possible winding arrangement ler draws a large current at slow speeds, typically
less than 0,2 per unit.
Soldering the wires to connection rings also

2.8
CONTROL SYSTEM
2.8 CONTROL SYSTEM
2.8.1 CAR CONTROL SYSTEM 2.8.1.2 Changes to the Car Control System in
2012
Based on the old teams experience and due to
2.8.1.1 Characteristics
the goal of a more energy efficient way of driv-
The car control system, CCS was designed in
ing, several changes have been done to the 2011
2009/2010 by Gunhndal (2010). It is a modular
CCS. Changes in the 2012 race requirments made
CAN-Bus based system. Different modules can be
additional changes nessesary too. All changes
easily added to or removed from the the system
can be distinguished by the topics below.
due to CAN-Bus technology.
Windshield Wiper
For the DNVFF in 2010 several modules was de-
The 2012 rules required a functionable elec-
signed to provide a broad range of functionality.
tric windshield wiper. A windshield wiper was
Some of the modules however were either not
designed by using a commercial front light wiper
functional or not nessesary for competing in the
engine that included a gearbox. This engine
battery electric class.
operates using 12V and 3A. Since the accessory
circuit uses a 24V battery an additional buck-
The DNVFF2 CCS uses a total of three modules.
converter circuit was nessesary in order to make
The front module handles the brake pedal sensor
the wiper engine run. Unfortunately there was no
input and operates the horn and front lights. The
free pin on the front module to include the wiper
driver interface is managed by the dashboard
functionality in the CCS. The buck-converter cir-
module. This module operates all button inputs
cuit was instead connected to be operated by a
and button LEDs as well as the driver display.
hardware switch totally independent of the CCS.
The engine module operates the rear lights and
communicates with the engine controller via RS-
232. It is the interface between the CCS and the
propulsional control system.

Two-Way-Communication setting the acceleration (speed ramp) and the
Based on last years experience it was crucial to starting routine were adjusted online.
get a working two way communication with the
SmartMotorController through the RS-232 inter- Automated Section Controlled Drive ASCD
face. With this in place the effects of all planed The biggest change in the CCS was the idea of
CCS changes and optimizations could be moni- having a optimized driving strategy for the Rot-
tored. The 2011 team put some effort into this terdam race track. This idea included an pre race
already but wasn´t able to make it work in 2011. track analysis to determine an optimized velocity
Starting off with the results achieved in 2011 a and torque profile around the track. The track
two-way-communication was established. Read- was devided into 4 different sections, each stert-
ing controller variables and parameters online ing after a 90° turn. The optimization was done
made a controller state observation possible. The by modelling the car and the track in MATLAB
communication rate between the engine control- and varying the speed, acceleration and coasting
ler and the engine module was not suffiient to times for each section. Wind and track inclina-
use the signals for any further online optimization were assumed to be zero (testing and racing
tion. Also a velocity integration in order to know showed that these influences must be considered
the distance traveled was impossible. By observ- for a more accurate optimization). The optimiza-
ing the engine controllers internal signals at high tion result was hardcoded in the CCS.
refresh rate using DSPComm a great measure-
ment inaccuracy was discovered. It can be stated The hardcoding of the optimization made CCS
that these measurements can not be used for any hardware changes nessesary. For each of the four
online optimization. Especially the DCLink power sections a dedicated button was included in the
signal shows a big variance. Therefore all optimi- CCS. Whenever a section button was pressed a
zation had to be done offline. hardcoded velocity profile was run. Therefore
A better understanding of the RS-232 commun- the two-way communication was necessesary.
cation lead to a more advanced communication Changing the old parameters for current speed
from the engine module to the engine controller.
For instance not just the speed reference variable
was changed while driving, but also variables

reference, the acceleration and the time when tor. The software running on the controller is
the car should coast determine the cars behavior also provided by the same sponsor. This means
till the next turn. This strategy of driving adds a that the team is not able to change the control-
optimality criteria and gives the driver the pos- ler code or to compile new code. However there
sibility to focus on other aspects of driving. are some parameters that can be changed via
the RS-232 interface. Since SmartMotor is a ex-
Since there are other cars racing on the track at traordanary cooperative sponsor, code changes
the same time, the hardcoded sections would have been done by them even in last minute.
have to be ended prematurely in certain cases. Compiled code can be flashed on the controller
Therefore a coast button was included in the CCS. by the team. The engine controller is a cascade
Whenever the driver senses a traffic jam situation linear controller for speed and torque.
the caost button is supposed to be pressed. Also
if wind pushes the car the driver can save energy As discribed in the CCS description parameters
by using this button. and variables can be changed and read via the
RS-232 interface by the CCSs engine module. This
To give feedback of the current engine controller gives the possibility to change the controllers
state, information such as the current speed and behavior in different situations.
the current speed reference is shown on the driv-
ers display at all times while driving. The engine used for racing was the same engine
as used in 2011. This engine has been designed
As a backup solution for racing and to make test- by Dahl-Jacobsen (2010). The batteries were
ing more efficient, a cruise control was imple- sponsored by Altitec and Gylling. For the batter-
mented in the CCS as well. The cruise control can ies charakteristics, see K-Brief Batteries.
be enabled by pressing the cruise control but-
ton on the steering wheel. Pressing this button 2.8.2.2 Changes to the Car Control System in
disables the ASCD section buttons by changing 2012
their functionality. The two section control but- Using a testbench at SmartMotor the drive trains
tons on the right now increase the current speed properties were determined especially for worst
reference. The two section control buttons on the case scenarios such as high load and low battery
left decrese the current speed reference. To be capacity. It became obvious that a under voltage
able to change the speed reference quickly, the limit is nessesary due to a high voltage drop for
upper buttons increase or decrease the speed low battery capacities. The testing also resulted
refernece in steps of 5 km/h respectively whereas in a adjustment of the engines airgap to prevent
the lower buttons are changing the speed refer- controller shut down due to a high back induced
ence by steps of 1 km/h. The coast button does voltage.
not change in functionality whenever the cruise
control is enabled. To prevent the battery or the controller to shut
down due to over load a torque reference limit
Using buttons for velocity changes made a gas was set including an anti wind up after the speed
pedal obsolete. The advantage of not having a controller.
gas pedal is a weight reduction and the fact that
it simplifies a energy efficient way of driving. In order to coast the contoller was set up with a
flystart routine that can be called via the RS-232
interface.
2.8.2 PROPULSIONAL CONTROL

SYSTEM
2.8.2.1 Characteristics
The propulsional control system defines the con-
trol system of the cars drive train. The main drive
train components are the battery, the load relay,
the engine controller and the electric engine.
The engine controller is sponsored by SmartMo-

Smart motor controller (SMC) converts DC-input to a three phase AC-signal driving the motor.
4 C
CONCLUSION
CAR PERFORMANCE team should contact professional manufacturers
early in the project regarding both molds and
production. A windshield wiper was successfully
The team is very satisified with the finished prod-
designed and implemented to satisfy the new
uct. The goal of a more robust, high quality and
SEM rule. The material used for the windshields
efficient car was achieved.
was easily scratched and a different material
should be used.
The carbon fiber monocoque is very good with
respect to weight, strength and aerodynamics. A
The change in power source from hydrogen to
new independent and fully damped suspension
battery was successful. The battery packs from
system worked as intended and together with
Gylling are very good and reliable. They deliver a
the strong monocoque enabled the driver to
very stable voltage all the way up to 17A which is
take the turns at full speed. A lot of effort was put
important for the SMC to work properly. They will
into the suspension geometry to elimate mis-
last several years if maintained properly and can
alignemnts and scrubbing of the wheels. Actually
be used for testing without wearing them out.
implementing this designed proved to be dif-
The battery pack from Altitec can deliver a lot
ficult. Laser equipment is needed to accurately
of power before shutting down (~35A), but will
align and adjust the suspension.
display a significant voltage drop when taken to
the limit.
Brakes are good and reliable, but lighter brake
discs can be looked into. Also the number of
The car control system (CCS) has a lot of helpful
degrees of freedom in the pivot point should be
functions for the driver, eliminating the need to
reduced. New hubs or improved spacers to elimi-
focus on smooth acceleration.
nate play and brake disc scrubbing.
Propulsion system needs to be totally re-thought.
Interior in general satisfied both driver comfort
The winning teams used separate motors for
and ergonomics, and looked pleasing for display-
coasting and acceleration. For startup it is better
ing the car. The steering wheel has good ergo-
to use smaller motors running at higher RPMs.
nomics, functionality and strength. Due to the
In 2008 and 2009 the NTNU team used a single
ASCS the driver was able to focus on positioning
small motor with good results. A hybrid between
the car on track and taking the turns in an opti-
the current- and old solution could be worth
mal way.
looking into. The motors effieciency is highly
influenced by the level of precision achieved
The extensive use of carbon fiber lead to chal-
during production. It is therefore highly recom-
lenges regarding load transfer and assembly.
mended to purchase the motors from a profes-
The super adhesive Araldite allowed us to secure
sional supplier.
parts of different materials to the carbon fiber
parts. However, the use of carbon fiber has some
The SMC driving the motors worked satisfactory.
limitations and challenges. The rims is an exam-
SmartMotor finally released the source code
ple of this and should be produced in another
making it easier for next years team to perform
material.
optimalizations. It is worth noting that the ef-
ficiency of the SMC is only about 90% and that
The production of the windows was very difficult.
more efficient controllers can be acquired.
Proper molds should be made out of a material
that can handle high temperatures. Next years

EXPERIENCES MADE DURING THE GENERAL
COMPETITION
The PR and media effort did not live up to the
During the competition in Rotterdam the team teams expectations. The media responsible
had some organizational problems. This made needs to be a member of the core team in order
it obvious that all team members should have to motivate the rest of the team.
clearly defined tasks. For test and race attempts
everyone needs to know where everyone else is, The systems engineers have developed a knowl-
who to contact, whom to not let out of sight etc. edge transfer system that will greatly aid next
A dedicated communication system should be years team. A3 sheets for each subsystem con-
acquired and tested before going to the race. This tains information necessary for next years team
would be very valuable to have, especially in case members to get a complete understanding of
of emergency. the car.
Maintenance routines should be made for critical

parts/systems. These need to be run through
before every run, whether testing or racing. A
proper maintenance plan could have prevented
the steering from failing on the last attempt in
Rotterdam. It was lucky that the previous attempt
was successful and nearly perfect. Kevlar in steer-
ing should be replaced to improve safety and
shorten maintenance intervals. Another principal
solution should be looked into. All lubricated
parts that are exposed to the elements should be
sealed or otherwise protected.
The black theme of the car and trailer caused

several problems. The matt black surface made
the inside of the car very hot which puts strain on
the driver. When presenting the project at DNV
in Høvik the car was put on display outside. The
heat from the sun made the back hatch, windows
and windshiled wiper loose their shape. Addi-
tionally, the increase in pressure in the front left
tire caused the rim to explode.

5
REFERENCES
REFERENCES
Alcoa. (2011). Fatigue Properties (confidential Granta Design Ltd. CES EduPack 2011. (n.d.).
text on Alumec fatigue properties obtained from
Alcoa). Gudem, M. (2006). Design and evaluation of sports
car suspension system. Trondheim: Norwegian
Beckman, B. (2008). The Physics of Racing. USA: University of Science and Technology.
Burbank.
Guhndal, A. (2010). Styre- og overvåkningssys-
(n.d.). Casting Resin RenCast. Solutions RenShape. tem for Shell Eco-Marathon kjøretøy. Trondheim,
Norway: Norwegian University of Science and
Chapman, C. B. (2002). A Risk Engineering Ap- Technology.
proach to Project Risk Management. International
Journal of Project Management , 5-16. Härkegård, G. (n.d.). Dimensjonering av maskinde-
ler. Trondheim: Tapir Akademiske Forlag.
Chapman, R. J. (1998). The Effectiveness of Working Irgens, F. (1999). Formelsamling mekanikk: sta-
Group Risk Identification and Assessment Tech- tikk, fasthetslære, dynamikk, fluidmekanikk (3
niques. International Journal of Project Manage- ed.). Trondheim: Tapir.
ment , 333-343.
Herroelen, W., & Leus, R. (2005). Project Scheduling
Dahl-Jacobsen, A. (2010). Energy Efficient Motor Under Uncertainty: Survey and Research Potentials.
for Shell Eco-marathon European Journal of Operational Research , 289-
Trondheim, Norway: Norwegian University of Sci- 306.
ence and Technology.
Herroelen, W., Demeulemeester, E., & Reyck, B. D.
Eikeland, H. A. & Lien, M. S. (2010). Aerodynamic (1997). A Classification Scheme for Project Schedul-
Development and Construction of a Car for Partici- ing Problems. Leuven: K.U.Leuven - Departement
pation in the Eco-marathon Competition. Trond- Toegepaste Economische Wetenschappen.
heim, Norway: Norwegian University of Science
and Technology. Huang, H., & Xu, C. (1998). Soft Budget Constraint
and the Optimal Choices of Research and Develop-
Endresen, F. V., Garmendia, I. Y., Gudvangen, ment Projects Financing. Journal of Comparative
H., Heidarloo, F. A., Larsen, P. T., Qviller, A., et al. Economics , 62-79.
(2011). Eco-marathon 2012 Project Report. Trond-
heim, Norway: Norwegian University of Science Jung, Y., & Woo, S. (2004). Flexible Work Breakdown
and Technology. Structure for Integrated Cost and Schedule Control.
Journal of Construction Engineering and Man-
Fastener Thread Designations and Definitions Pitch, agement , 616-625.
Minor, Major Diameters. (n.d.). Retrieved Febru-
ary 2012, from http://www.engineersedge.com/ Kaplan, S., & Garrick, B. J. (1981). On The Quantita-
hardware/metric-external-thread-sizes1.htm tive Definition of Risk. Risk Analysis , 11-27.
Garcı´a-Fornieles, J. M., Fan, I. S., Perez, A., Wain- Kolisch, R., & Padman, R. (2001). An Integrated
wright, C., & Sehdev, K. (2003). A Work Breakdown Survey of Deterministic Project Scheduling. Omega-
Structure that Integrates Different Views in Aircraft The International Journal of Science , 249-272.
Modification Projects. Concurrent Engineering ,
47-54. Tonning, O. (2012). Implementing Lean System
Engineering in the DNV Fuel Fighter project. Trond-
Gillespie, T. (1992). Fundamentals of Vehicle Dy- heim, Norway: Norwegian University of Science
namics. USA: Society of Automotive Engineers, and Technology.
Inc.
PMI. (2008). A Guide to the Project Management

Body of Knowledge. Project Management Insti-
tute, Inc.
Proceedings of the Institution of Mechanical Engi-

neers: Vehicle Ride and Handling. (n.d.).
Qviller, A., Stockfleth, T., Bleie, S., & Hoel, M.
(2011). Shell Eco-marathon, DNV Fuel Fighter
2011. Trondheim: NTNU.
Rølvåg, T. (2008). Design and Optimization of

Suspension Systems and Component. Trondheim:
Department of Machine Design and Materials
Technology, Norwegian University of Science and
Technology.
SAE. (2012). Bosch Automotive Handbook.
Shell. (n.d.). Ahoy race track map. Retrieved No-

vember 2011, from http://www.shell.com/home/
content/ecomarathon/europe/2012_rotterdam/
competition_structure
Shell. (n.d.). Chapter I Official Rules. Retrieved No-

vember 2011, from Chapter I Official Rules: http://
www.shell.com/home/content/ecomarathon/
europe/for_participants/europe_rules/
SKF. (n.d.). Selection of bearing size. Retrieved Feb-

ruary 2012, from Selection of bearing size.
Tatikonda, M. V., & Rosenthal, S. R. (2000). Tech-

nology Novelty, Project Complexity, and Product
Development Project Execution Success: A Deeper
Look at Task Uncertainty in Product Innovation.
IEEE Transactions on Engineering Management ,
74-87.
Vedvik, N. (2011). Composite Materials. Trond-

heim: Norwegian University of Science and
Technology.
Venkataraman, R. R., & Pinto, J. K. (2008). Cost and

Value Management in Projects. John Wiley & Sons,
Inc.
Yuguero-Garmendia, I. (2012). Verification, Valida-

tion and Testing Activities of the DNV Fuel Figheter
2. Trondheim, Norway: Norwegian University of
Science and Technology.

LIST OF FIGURES
Figure 2.1.1 Work breakdown structure 23
Figure 2.1.2 Milestones arranged in the schedule 25
Figure 2.1.3 Planned S-curve of the project 26
Figure 2.1.4 Detailed budget 29
Figure 2.2.1 The template uses a Plan-Do-Check-Act approach, meaning the problem is
solved through analysis, testing, verification and then implemented 36
Figure 2.2.2 The front page is dedicated to listing components, interface diagrams, describ-
ing major decisions and the manufacturing process. The components are cat-
egorized as ‘New Product Development’ (NPD) meaning designed by this year’s
team, ‘Reused’ (R) meaning an unaltered design or ‘Purchased’ (P) meaning
bought from a manufacturer. Also, the components are rated for satisfaction
stating how well the component behaved relative to its expected behaviour 37
Figure 2.2.3 Page 2 is dedicated to the 3D-models, simulations and analyzes 38
Figure 2.2.4 Page 3 is dedicated to perceived risks, a description of problems and how they 39
were solved, and proposed future work
Figure 2.2.5 The sub-system “Wheel” broken down into its components 40
Figure 2.2.6 The elements that make up the wheel are positioned along the diagonal. Lines
between the elements show interfaces between them (emphasized by dia-
monds). The Hub is the part that links the sub-system to the rest of the system 40
Figure 2.2.8 VPB for DNVFF2. Tasks are numbered on the right-hand side. The Gantt-dia-
gram in the middle shows what day a certain task is due. Colors tell whether a
task has slipped or not. The left-hand side is for information and exchanging
messages 41
Figure 2.2.9 Risk cube display 42
Figure 2.2.10 Curve showing how average risk level developed through the project 42
Figure 2.2.11 Variation of program risk and effort throughout system development (Kosia-
koff et al, 2011) 43
Figure 2.2.12 Template for the Timeline 43
Figure 2.2.13 Template for the Wall-Architecture. Orange means “in production”, red is “criti-
cal”, yellow is “pending”, green is “produced” and blue is “assembled” 44
Figure 2.4.1 Map of the Ahoy. The track is highlighted in red. The numbers indicate corners
and driving direction 59
Figure 3.1.1 Suspension system modeled with RB2 elements, loads and moments are
applied in the wheel center and transferred into the body through the stiff
suspension system. 73
Figure 3.1.2 Displacement due to loads during race conditions 74
Figure 3.1.3 Displacement due to loads during testing 74
Figure 3.1.4 Layup for the car, placement of 4 layers (top), core reinforcemenst (middle) and
2 layers (bottom) 75
Figure 3.1.5 Stresses in the most highly stressed carbon fiber ply 76
Figure 3.1.6 A draft angle analysis done in Catia to discover what needed to be improved
before the production 78
Figure 3.1.7 3D model of molds 79
Figure 3.1.8 Overlay in a corner, two layer layup leads to four layers in the splice area 86
Figure 3.1.9 Overview of the final layup of the car 89
Figure 3.1.10 Arrangement of the windows 108
Figure 3.1.11 The holes for the lights needed to be planned before the production of the
monocoque started, and therefore also before the inside of the lights were
designed 113
Figure 3.1.12 Different combinations of light placement were tested in SolidWorks 115
Figure 3.1.13 The final front light and rear light concepts 116
Figure 3.1.14 Different placements of the hinges 120
Figure 3.1.15 Different door opening solutions 120
Figure 3.1.16 Idea generation of different hinge concepts 121
Figure 3.1.17 Idea generation of different door handle concepts 123
Figure 3.1.18 Principle drawing of the hinge showing the opening motion 122

Figure 3.1.19 Door handle in the open postion 123
Figure 3.1.20 Idea generation of windshield wiper 124
Figure 3.1.21 Different wiper motions 124
Figure 3.1.22 Idea generation for mounting the back hatch 129
Figure 3.1.23 Idea generation for mounting the side covers 130
Figure 3.1.24 Idea generation for the towing hook 132
Figure 3.1.25 Strength analysis of the towing hook. The 2000 N load yield maximum stress of
110 MPa 132
Figure 3.2.1 Before the design of the interior could start, one needed to establish a layout
of the cockpit, and in that way set an outline for the size of each component 152
Figure 3.2.2 An 3D-model with the outlines for the dashboard were given to group
working with idea generation 154
Figure 3.2.3 Concept 1 had a removable top cover to provide easy access 156
Figure 3.2.4 Concept 2 consisted of a light construction attached to both the monocoque
shell and the floor. It had no removeable parts 156
Figure 3.2.5 3D-model of front beam mold ready for milling. The mold is milled in two parts
and then glued together later 157
Figure 3.2.6 Skething was done to explore ideas, but most of the ideation process was
done with physical models and 3D-models 166
Figure 3.2.7 Concept 1 168
Figure 3.2.10 Criteria matrix 171
Figure 3.2.11 The molds for the vacuum forming was first designed in SolidWorks and then
milled out in a CNC milling machine 172
Figure 3.2.12 Seat Interfaces 177
Figure 3.2.13 Basic Ergonomics 178
Figure 3.2.14 A lot of explorative sketching was done to generate as many ideas as possible 180
Figure 3.2.15 3D-modelling was used to test different ideas and to decide on different di-
mensions in a fast way 181
Figure 3.2.16 Idea 1 182
Figure 3.2.19 All the different seat ideas, where idea number 2 was selected for further de-
velopment 185
Figure 3.2.20 Details of the final concept , cushions added 186
Figure 3.2.21 Detailed model of the interior without mirrors 190
Figure 3.2.22 Explorative sketching was done to generate ideas 191
Figure 3.2.23 The three different concepts seen from the drivers view. Closeup of the differ-
ent concepts are shown below 192
Figure 3.2.24 The main part for the mirrors designed in SolidWorks. The model was planned
in a way that made it easy to mill 193
Figure 3.3.1 Typical steering knuckle concept 196
Figure 3.3.2 Double wishbone system, parts annotated 197
Figure 3.3.3 Camber angle change during spring action 198
Figure 3.3.4 Top view front view of front left suspension links 198
Figure 3.3.5 Viewed from beneath, front right suspensions 198
Figure 3.3.6 Rear view of front left suspension links 198
Figure 3.3.7 Rigid body model of front suspension, where PB and PC are the braking and cor-
nering forces, and C1, 2, 3, 4 are the support forces acting through the wishbone
links 199
Figure 3.3.8 Force and constraints on the knuckle 199
Figure 3.3.9 Simplified force and torque on knuckle 199
Figure 3.3.10 Steering knuckle modeled as a beam with load and reaction forces 200
Figure 3.3.11 Constraint model for top arms during braking 200
Figure 3.3.12 Constraint model for lower arms during braking 200

Figure 3.3.13 Constraint model for top arms during cornering 201
Figure 3.3.14 Constraint model for lower arms during cornering 201
Figure 3.3.15 Diagram of weight force and link
geometry view from behind the car, where PW is the weight on the wheel and
C5 is the coilover’s reaction force 202
Figure 3.3.16 Sketch setup in NX to obtain measurements for anti-dive calculations 202
Figure 3.3.17 Force diagram of dive/anti-dive contributing forces 203
Figure 3.3.18 Front and back view of the hub 204
Figure 3.3.19 View of hub assembly 204
Figure 3.3.20 FEM analysis of hub during braking 204
Figure 3.3.21 FEM analysis of hub during cornering 204
Figure 3.3.22 CAD model of axle 205
Figure 3.3.23 FEM analysis of axle during braking 205
Figure 3.3.24 Ackerman steering based on sketch in the PD-journal, wheels and angles an-
notated. α is the typical toe-angle when driving straight ahead. 206
Figure 3.3.25 View of steering radius check 207
Figure 3.3.26 Verification of ackerman steering for 20 m turning radius 207
Figure 3.3.27 Verification of 6 m turning radius requirement 208
Figure 3.3.28 Finding the smallest possible turning radius, tilted view 208
Figure 3.3.29 FEM analysis of knuckle during braking 208
Figure 3.3.30 FEM analysis of knuckle during cornering 208
Figure 3.3.31 Unidirectional carbon fiber wrapping for reinforcement of bolted connections 209
Figure 3.3.32 Mold for carbon fiber knuckles 209
Figure 3.3.33 Link consisting of rod ends and nuts 210
Figure 3.3.34 Single rod end on bottom, double rod ends on top 211
Figure 3.3.35 Solid lower wishbone 211
Figure 3.3.36 Connector and link rods 211
Figure 3.3.37 3D view of the left wishbone connector 211
Figure 3.3.38 Left suspension with connector and coilover 212
Figure 3.3.39 FEM analysis of wishbone connector during braking 212
Figure 3.3.40 Tie rod assembly 214
Figure 3.3.41 Left tie rod on suspension seen from beneath 214
Figure 3.3.42 Front left wheel well on the final body shell 214
Figure 3.3.43 The final clevis design. Purple highlights the vibration damping bushings 215
Figure 3.3.44 FEM analysis of clevis 215
Figure 3.3.45 Clevises attached to wheel well using wedges and bushings 215
Figure 3.3.46 3D CAD model of the rear suspension developed during the spring 2012 216
Figure 3.3.47 Knuckle 216
Figure 3.3.48 High vertical distance between the ground and the bottom of the rear wheel
well wall 216
Figure 3.3.49 Distance between the lower rod ends and the brake disc increased on the new
design 217
Figure 3.3.50 How to determine the roll center of an independent suspension 218
Figure 3.3.51 Mounting points for A-arm connector and toe link on the same horizontal line
to prevent bump- and roll steer 218
Figure 3.3.52 2D-model of the suspensions front-to-back geometry in UGS NX 7.5. The most
important dimensions are shown in the illustration. 219
Figure 3.3.53 FE analysis setup for the knuckle
Figure 3.3.54 Test result from the final analysis of the knuckle 219
Figure 3.3.55 STEP - model of the knuckle imported into NX IDEAS 5.0. The couloured lines
show the milling path 220
Figure 3.3.56 3D model of the mold and the knuckle 220
Figure 3.3.57 Lower A-arm connector 222
Figure 3.3.58 Forces acting on the suspension 222
Figure 3.3.59 Forces acting on the connector 223
Figure 3.3.60 FE analysis setup for the connector 223

Figure 3.3.61 Test result from the final analysis of the knuckle 224
Figure 3.3.62 Coilover cross section overview 224
Figure 3.3.63 Determining spring travel and spring force 225
Figure 3.3.64 FE analysis of the upper spring holder 226
Figure 3.3.65 FE analysis of the lower spring holder 227
Figure 3.3.66 FE analysis of the piston 227
Figure 3.3.67 Toe link extension 227
Figure 3.3.68 Rods and rod ends 228
Figure 3.4.1 All steering components in the final assembly 235
Figure 3.4.2 Length adjustment concept 235
Figure 3.4.3 Drag link length adjustment 236
Figure 3.4.4 Old tie rod solution 236
Figure 3.4.5 Tie rod solution chosen 236
Figure 3.4.6 PU bearing, provides low friction for the drag link 237
Figure 3.4.7 3D model of the steering column, with sub parts assembled 237
Figure 3.4.8 Thread house 238
Figure 3.4.9 Steering wheel mounting socket 238
Figure 3.4.10 Pulley assembled on bracket 239
Figure 3.4.11 Pulley assembled on bracket 239
Figure 3.4.12 Strength analysis of the steering column, maximum stress of 69 MPa 239
Figure 3.5.1 Exploded view of the brake pedal 247
Figure 3.5.2 Stages of the brake pedal design, from early concept to final design 248
Figure 3.5.3 Pivot point force sketch 248
Figure 3.5.4 The maximum stress of 268 MPa was located around the pivot point hole,
which yields a safety factor of 2. The analysis showed a deflection of 2.5 mm,
which will be a conservatiove results due to the material data 249
Figure 3.5.5 Placing of master cylinders and pedal arm. From this the minimum distance
between components was determined 249
Figure 3.5.6 Grooves removed from block and final piece 250
Figure 3.5.7 The maximum stress was 136 MPa, giving a safety factor of 4. This stress oc-
curred at the pedal arm support. There were no significantly deformations in
the pedal box 251
Figure 3.5.8 Displacements in the pedal box 251
Figure 3.6.1 Exploded 3D CAD model of the rims produced in 2012 256
Figure 3.6.2 FE analysis set up of the rims 256
Figure 3.6.3 FE analysis results of the rims 256
Figure 3.7.1 Exploded view of comlpeted motor assembly 262
Figure 3.7.2 Cross section of motor 263
Figure 3.7.3 Close up of the motor hub 264
Figure 3.7.4 3D view of the new motor axle 264
Figure 3.7.5 Load cases for an axle during cornering 264
Figure 3.7.6 FEM analysis of motor axle subject to a hard bump 264
Figure 3.7.7 FEM analysis of motor axle on inner wheel of a typical, sharp curve 265
Figure 3.7.8 FEM analysis of motor axle on outer wheel of a typical, sharp curve 265
Figure 3.7.9 Outer and inner rotor with colorized magnet arrays 265
Figure 3.7.10 Close up of rim adapters, in green and blue. The yellow section is the rotor
spacer 266
Figure 3.7.11 The new test adapter 266
Figure 3.7.12 Magnets in a Hallbach array, arrows showing direction of magnetization 268
Figure 3.7.13 Magnets and iron ring, arrows showing direction of magnetization 268
Figure 3.7.14 Single phase single layer wave winding 268
Figure 3.7.15 Three phase single layer wave winding 268
Figure 3.7.16 Single phase distributed winding 268
Figure 3.7.17 Three phase distributed winding 268
Figure 3.7.18 Winding connection 276

Figure 3.7.19 Efficiency as a function of torque of the old engine at 300 rpm, 35, 55 km/h as
it was in 2011 278
Figure 3.7.20 Efficiency as a function of torque of the new engine at 257,5 rpm, 26 km/h with
the smallest obtainable air gap 278
Figure 3.7.21 Possible winding arrangement 279

LIST OF TABLES
Table 2.1.1 Improvement points for developing a new vehicle 20
Table 2.1.2 Weighting of different elements 26
Table 2.1.3 Detailed budget 29
Table 2.1.4 Definition of rating system 31
Table 3.1.1 Input data for calculations of lift and drag 143
Table 3.1.2 Results from wind tunnel testing 143
Table 3.3.1 Trade-off matrix for steering knuckle material selection 206
Table 3.3.2 Table of minimum pitch and minor diameters of ISO M6 and M8 210
Table 3.3.3 FE analysis set up for the knuckle. DOF 1-3 is displacement in x-, y-, and z-direc-
tion, respectively. DOF 4-6 is rotation about x-, y-, and z-axis, respectively 219
Table 3.3.4 FE analysis set up and result for the knuckle 219
Table 3.3.5 FE analysis set up for the knuckle 223
Table 3.3.6 FE analysis set up and result for the connector 224
Table 3.3.7 Changes in spring rate 225
Table 3.3.8 Properties of the SF-TF 1901 compression spring ordered from Lesjöfors (Lesjö-
fors, 2012) 226
Table 3.3.9 FE analysis of the upper spring holder 226
Table 3.3.10 FE analysis of the lower spring holder 227
Table 3.3.11 FE analysis of the piston 227
Table 3.6.1 FE analysis set up for the rims 256
Table 3.7.1 Requirements specification for mechanical motor parts 263

APPENDIX
Appendix A: Acknowledgments
I would like to thank the following people for their support, either directly or indirectly, to the progress and
success of this project:
• Knut Einar Aasland

IPM staff, NTNU
Supervisor
• Kristina Dalberg
Det Norske Veritas (DNV)
Main sponsor contact person
• Bjarne Stolpnessæter
IPM staff, NTNU
For all the CNC milling of moulds, suspension knuckles and a lot more, for always being open to new ma-
chining requests and short delivery times
• Per Øystein Nortug

IPM workshop, NTNU
For machining assistance and staying late nights, even weekends, so we could work in the workshop
• Børge Holen
IPM workshop, NTNU
For machining assistance
• Office Adminstration
IPM staff, NTNU
For handling all the purchasing orders and payments
• Arne Gellein
SINTEF Materialer og Kjemi
For free metal cutting
• Jan Erik Molde

Elkraft, NTNU
For machining of motor hub

Appendix B: Estimated material data for DB420
Materia type: Orthotropic
Min Max Units

Youngs modulus E1 110 140 GPa
E2 5 8 GPa
E3 GPa
Poissons ratio NU12 0.25 0.30 -
NU23 0.50 0.60 -
NU13 -
Shear modulus G12 3 5 GPa
G23 GPa
G13 GPa
Tensile stress ST1 1500 2000 MPa
ST2 MPa
ST3 20 50 MPa
Compressive stress SC1 800 1200 MPa
SC2 MPa
SC3 100 200 MPa
Shear stress S12 40 80 MPa
S23 MPa
S13 MPa
Density 420 g/m2
Density, cured 1.5 g/cm3
Thickness 0.8 mm
Thickness, cured 0.5 mm
NB: DB420 er en strikket armering bestående av 2 distinkte lag, det er IKKE en vev. Altså gir en ammering en
layup [45/-45]. Tykkelse på hvert lag er jo gitt av tettheter og volumfraksjoner, så det finner du greit.
Angående volumfraksjon, vil en rimelig range være 0.50-0.55 med godt vakuum. Slurv og fanteri forøvrig kan
gi lavere, f.eks. 0.45.

Appendix C: Program listing
Program: Main
O0150 ; Main program for milling front left wishbone
S3000 M03 F1400 G17 G90 ;
M08 ; Coolant on
M21 ; Optional: Mirror X-axis, for right wishbone
M22 ; Optional: Mirror Y-axis, for right wishbone
M23 ; Optional: Mirror OFF
G00 X-15. Y0. ; Move outside material block
Z-90. ;
G01 Z-103. ; Prepare cut depth for first cut
M98 P151 L7 ; Cut seven times with 3mm cuts
G90 ; Switch back to absolute coordinates just in case
G01 Z-123. ; Prepare height for final planar cut
M98 P151 L1 ; Last planar cut
N11 G90 ; Labeled for graph plot
G01 X-15. ;
G00 Z-90. ;
X40. Y-28.8 ;
G01 Z-106. ;
X35. ;
M98 P152 L21 ; Mill north ear profile
G90 ;
G00 Z-90. ;
X40. Y-45.8 ;
G01 Z-106. ;
X35. ;
M98 P152 L21 ; Mill south ear profile
G90 ;
G00 Z-90. ;
X-15. ;
Y-59.1 ;
G01 Z-125. ;
M98 P153 L8 ; Cut out the bottom angles
G90 ;
G01 Z-90. ; Begin safe return to origin
M09 ; Coolant off
M05 ; Spindle stop
G00 Z0. ; To origin
X0. Y0. ;
M30 ; Program end/all reset
%

Program: Rough cut
O0151 ; Program for milling around ears until Z-middle of the wishbone
Assumes starting in position X-15. Y0.
G17 G90 ;
G01 X0. ;
Y-41.9 ;
X30. ;
G03 X35. Y-36.9 R5. ;
G01 Y-30.9 ;
G03 X30. Y-25.9 R5. ;
G01 X0. ;
Y-58.9 ;
X30. ;
G03 X35. Y-53.9 R5. ;
G01 Y-47.9 ;
G03 X30. Y-42.9 R5. ;
G01 X0. ;
Y-63.1 ; To south edge
X61.3 ; To south-east vertex
Y0. ; To north-east vertex
X5. ; To second west edge
Y-20.9 ; To second north edge of north ear
X40. ; To second east edge of both ears
Y-58.1 ; To second south edge
X56.3 ; To second east edge
Y-5. ; To second north edge
X10. ; To third west edge
Y-15.9 ; To third north edge of north ear
X45. ;
Y-53.1 ;
X50. ;
Y-10. ;
X-15. ;
Y0. ;
G91 ;
G01 Z-3. ; Lower for next cut
G90 ;
M99 ;
%

Program: Ear profile
O0152 ; Program for milling ear profiles, operates only in ZX-plane, caller must modify
Y-axis
Assumes starting in position X35. Z-106. and Y depending on which is being
milled
G18 G90 ; Select ZX-plane for circular interpolation and abs. coords.
G02 X26.5 Z-100. R9.021 ; East radius
G01 X16.2 ;
G02 X8.586 Z-104.739 R8.501 ; West radius
G01 X0. Z-121.910 ; West slope
G91 ;
G01 Y-0.2 ; Strafe for next cut
G90 ;
G01 X8.586 Z-104.739 ; West slope
G03 X16.2 Z-100. R8.501 ; West radius
G01 X26.5 ;
G03 X35. Z-106. R9.021 ; East radius
G91
G01 Y-0.2 ; Strafe for next cut
G90 ;
M99 ;
%
Program: Final shape

O0153 ; Program for milling angle profile
Assumes starting in position X-15. Y-59.1
G17 G90 ; Select XY-plane for circular interpolation and abs. coords.
G01 X0. ;
Y-28. ;
X34.7 Y0. ;
X52. Y-20.6 ;
X42.1 Y-28.9 ;
G03 X43.6 Y-34.2 R3. ;
G01 X61.3 Y-36.7 ;
X57.6 Y63.1 ;
X0. Y-59.1 ;
X-5. ;
G91 ;
G01 Z-2. ; Lower spindle for the next cut
G90 ;
M99 ;
%

Appendix D: Program listing
Program: Main
O0150 ; Main program for milling front left wishbone
S3000 M03 F1400 G17 G90 ;
M08 ; Coolant on
M21 ; Optional: Mirror X-axis, for right wishbone
M22 ; Optional: Mirror Y-axis, for right wishbone
M23 ; Optional: Mirror OFF
G00 X-15. Y0. ; Move outside material block
Z-90. ;
G01 Z-103. ; Prepare cut depth for first cut
M98 P151 L7 ; Cut seven times with 3mm cuts
G90 ; Switch back to absolute coordinates just in case
G01 Z-123. ; Prepare height for final planar cut
M98 P151 L1 ; Last planar cut
N11 G90 ; Labeled for graph plot
G01 X-15. ;
G00 Z-90. ;
X40. Y-28.8 ;
G01 Z-106. ;
X35. ;
M98 P152 L21 ; Mill north ear profile
G90 ;
G00 Z-90. ;
X40. Y-45.8 ;
G01 Z-106. ;
X35. ;
M98 P152 L21 ; Mill south ear profile
G90 ;
G00 Z-90. ;
X-15. ;
Y-59.1 ;
G01 Z-125. ;
M98 P153 L8 ; Cut out the bottom angles
G90 ;
G01 Z-90. ; Begin safe return to origin
M09 ; Coolant off
M05 ; Spindle stop
G00 Z0. ; To origin
X0. Y0. ;
M30 ; Program end/all reset
%

Appendix E: Spreadsheet for simplified vehicle dynamics

Appendix F: Bolt-carbon fiber tensile tests
Bolt-carbon fiber tensile tests

NTNU
Hans Gudvangen
Aksel Qviller
February 2012
The goal of the testing was to verify the viability of using a bolted connection to massive carbon fiber
for our car’s suspension. The size of the bolt was dictated by the size of the end rods that we were
aiming to use. The test setup therefore consisted of a short rod of massive carbon fiber with two M8
bolts inserted into either end. The end holes were carefully drilled to Ø6.8 mm and thread-tapped to
M8. Two bolts were inserted through connectors and screwed into the CF test piece, then pulled
until failure in a tensile testing machine.

The CF rods were sawed off from a larger piece of massive CF. The first tests quickly revealed that the
CF would delaminate around one of the bolt holes.
We proposed to reinforce the ends with layers of CF UD wrapped around. For these tests, this would
be unpractical, so we used two clamps from aluminum and M4 bolts. These were tightened to
simulate the additional stiffness provided by wrapping UD layers around the ends of the CF rod.
Additionally, we tried to glue the bolts into the holes using Araldite 2031, both with and without
threads in the hole of the CF piece.

The actual test setup is illustrated below:
The results from the different tests, although few, were consistent and is summarized as follows:
Thread dimension M8x1,25

Hole depth 20 mm
Thread depth 15 mm
Result (F1 max)

Test 1 – threads without glue 7,5 [kN]
Test 2 – threads with Araldite 2031 9,0 [kN]
Test 3 – 8,5 mm hole 20 mm deep with Araldite 2031 9,0 [kN]

312
Energy source Benni Gutjahr, 2012
BLUE BATTERY WHITE BATTERY
ECO-MARATHON 2012
Appendix G: K-brief example
Specifications:
- sponsor: Altitec
- battery charger included
- nom voltage 48V
- nom discharge current 24A
- capacity 5800mAh Specifications:
- sponsor: Gylling
- battery charger NOT included
- nom voltage 46,2V
Problems : - max discharge current 19A
- for power drawing >300W depending on discharge capacity very high voltage drop - capacity 4400mAh
- voltage drop often triggers the BMSs under voltage protection and shuts down battery
- battery is likely to get warm which triggered BMSs temperature protection one time while testing Battery is charged by a power source (rating: 60V/10A) from the Electrical department.
Conclusion:
- battery is not suitable for race conditions in combination with the 2012 drive drain
Problems:
Charger: - limited current discharge ability
- BMS over current prtection triggers latest at 19A
- SmartMotor engine controllers fly start routine at high torque (>12Nm) not possible
- Limited torque for drive train due to limited current
Conclusion:
- battery is of limited suitability for the 2012 drive drain
ASSEMBLY TEST:
Subsystem: S.8 Brake circuit Responsible:H.J.S Test Date: 19.04.2012 Place: Dragvoll Sport center
Test Test ID. Procedure Requirement for passing the Personnel Material/ Performance Comments Status Correcting
no. test Tools (measures) Pass/No actions
pass
1 Effectiveness of Place the car on a The vehicle must remain Driver and team member The car must remain The car remained Pass
the braking circuit 20%slope. stationary. pushing the car stationary stationary and then it was
also pushed while the
Appendix G: Assembly test example
braking.
2 Braking Move the vehicle and test Enough brake force Driver and team member Pass
capabilities the brakes pushing the car
3 Leaks Press hard on brake pedal No visible leaks Driver and team member Entire No visible leaks have Pass
to check if there are some Brake to appear
leaks system
4 Brake pedal Measure the dimensions Minimum surface area of brake One team member to The Brake pedal is Pass
dimensions pedal: 25 cm2. measure it approximately 49 cm2
5 Braking Combinations of weak, The brakes must be strong Driver and another team The brake systems Pass
effectiveness gradual, hard and instant enough and have to stop the car member to push the car must be reliable and
braking as soon as the driver steps on and to measure the able to perform
the pedal distance
with low and higher
speed, as well as during
turning at different radius.
Braking:
Initial braking should be cautious to listen to how the brake pads attack the discs, and the vibrations that are transferred to the chassis and amplified.
We should try combinations of weak, gradual, hard and instant braking with low and higher speed, as well as during turning at different radius.
We will also need to check whether the rear wheels begin to skid (if allowed by the gym center) - if they do, too much force is routed to the rear brakes.
ECO-MARATHON 2012
313

Ntnu 2012 PDF

Uploaded by

Copyright:

Available Formats

Ntnu 2012 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ntnu 2012 PDF

Uploaded by

Copyright:

Available Formats

Development and Construction of Vehicle

for Participation in the Shell

Aslak Brage Espeland

Product Design and Manufacturing

Norwegian University of Science and Technology

Trondhiem, June 2012

Aslak Brage Espeland Hans Gudvangen

Håkon Johan Seiness Petter Thorrud Larsen

Name of part Major Improvement Points Scope of Impact on Making

Table 2.1.1: Improvement points for developing a new vehicle

Figure 2.1.1: Work breakdown structure

Figure 2.1.2: Milestones arranged in the schedule

Figure 2.1.3: Planned S-curve of the project

Figure 2.1.4: Detailed budget

Scale Probability of Occurrence/Impact Related Color

Table 2.1.4: Definition of rating system

Another way of standardizing the process is to

Figure 2.2.1 shows a template for a K-brief that

The following three sides show a three-page

2.2.1.2 Storing and Sharing Knowledge with

Figure 2.2.6 shows how an N2 diagram may be

2.2.2.2 Visualizing Project Progress

The Timeline divides the project into one-week

Figure 2.2.12: Template for the Timeline

SEM 2012 team preparing the visibility check

DNV FUEL FIGHTER 2

DNV FUEL FIGHTER 2

Wordpress was chosen as the Content

The main event in the spring semester was

Image taken from: http://static.zoom.nl/ 10.06.12

Figure 3.1.2: Displacement due to loads during race conditions

Figure 3.1.3: Displacement due to loads during testing

3.1.2.3 Results solution can be to make holes in the core mate-

3.1.3 DESIGNING THE MOLDS

Indents for windows

Lines around lights

Flanges around the edge Lines around the door

Figure 3.1.7: 3D model of molds

To finalise the mold for production, horizontal

Bottom mold after fine milling

Milling of epoxy paste

3.1.4.2 Layup lap of 40 – 60 mm, see figure 3.1.8. By using this

The first step was to lay a line of vacuum sealer

Carbon fiber laid with an overlay of total 40 – 60 mm.

Figure 3.1.9: Overview of the final layup of the car

3.1.4.3 Doors 3.1.4.4 Firewall

Production of the doors follow the exact same

Layup of the side door

Rear wheel well covers

Water reservoir method used to produce rear wheel well covers

Gluing of the side parts of the front wheel well covers

100 ECO-MARATHON 2012

3.1.4.3 Doors resulted in softening the carbon fiber and the

102 ECO-MARATHON 2012

Rear wheel covers fitted to the monocoque

104 ECO-MARATHON 2012

Sanding of the whole car, a time consuming process

106 ECO-MARATHON 2012

ECO-MARATHON 2012 107