Vehicle Engineering

VEHICLE
ENGINEERING
www.AcademyPublish.org
Vehicle Engineering
Other access free resources from AcademyPublish.org
Journal of Engineering and Technology,

academypublish.org/journal/show/title/journal-of-engineering-and-technology
The journal covers the latest developments and research in the fields of engineering,
technology and applied sciences. The aim is to publish papers that cover wide range of
topics promoting theoretical and experimental engineering.
Open Journal of Economic Research,

academypublish.org/journal/show/title/open-journal-of-economic-research
It tends to publishpapers covering all branches of conceptual, theoretical and practical

economics and wide range of other fields it coats.This journal aims tosignificantly contribute
to progress and understanding of how economical forces allocate scarce resources and
services.
Journal of Computer Sciences and Information Technology,

www.academypublish.org/journal/show/title/journal-of-computer-and-information-technology
Journals heterogeneous scope reaches extended topics of theoretical approach to computer

science that also provide new insights to commercial, social and educational values of software
and information technology.
Journal of Medical Research and Science,

academypublish.org/journal/show/title/journal-of-medical-research-and-science
One of the goals is to regularly provide plurality of topics in each issue. By dispersing the focal
point, the Journal will be able to publish most recent and most sensational papers, making it a
reliable and permanently up-to-date.
Vehicle Engineering
Vehicle Engineering
Published by:
AcademyPublish.org (Publishing Services LLC), 2120 CareyAvenue, Cheyenne, WY 82001, USA
AcademyPublish.org is an open access publisher that provides all publication at homepage

access free to all internet users. For more information please contact us at:
contact@academypublish.org
Cover Design: AP staff

Interior: AP staff
Printing History: February 2014: First edition.
Publishing Services LLC (and logo design) and Academy Publish (and logo design)
are trademarks of Publishing Services LLC. Any use of Publishing Services LLC and
AcademyPublish.org trademarks without written permission from the company is strictly
prohibited.
All material (chapters) published in this book are fully open access, meaning they are
immediately freely available to read, copy and share to all readers. The rights are not
transferred to AcademyPublish, that is, authors retain all rights. AcademyPublish.orgs
Open Access articles and publications are available at www.academypublish.org access
free.
You understand and agree that Publishing Services LLC does not endorse or make any
representations as to the reliability or accuracy of any author generated content published
in this publication. While every precaution has been taken in the preparation of this book,
the publisher and authors assume no responsibility for errors or omissions, or for damages
resulting from the use of the information contained herein. You are fully responsible for
your own use of such content.
ISBN: 978-0-9835850-8-4
About the Authors
Section 1 - New Technologies and Advances in Automotive Engineering
The Current Status of Automotive Coatings and Future Trends Using Nanotechnology,
p.21-p.51
Dr. Mohsen Mohseni, H. Yari, B. Ramezanzadeh
Dr. Mohsen Mohseni

Associate Professor
Department of Polymer Eng. and Color Tech.
Amirkabir University of Technology
424 Hafez St.
P.O.Box 15875-4414
Tehran- Iran
Tel: 0098-21-64542441
Fax: 0098-21-66469162
e-mail: mmohseni@aut.ac.ir
website: www.aut.ac.ir\mmohseni
Automotive Lower Suspension Arm: Experimental and Simulation Durability

Assessment, p.53-p.74
Dr. Kadhim obtained the PhD degree in 2011 from the Department of Mechanical and
aspiration for success, she joined the university as a Postdoctoral fellow. Her research
skills helped to form the foundation and component life prediction philosophy that she
applies to fatigue analysis. She has valuable hands on experimental and analytical fatigue
assessment. In addition, like academic and applied research, fatigue analysis requires
global thinking and a panoramic approach to problem solving. This type of analysis is
fundamental research to any industry like automotive, aerospace, marine, etc. Therefore,
analysis as thoroughly as possible, then achieve the component life optimization. She
seminar proceedings.
Contact: nawar2007@gmail.com
Professor Nikabdullah obtained the PhD in 1991 from The Technical University of Berlin.
with Universiti Kebangsaan Malaysia (UKM) since February 1992, after servicing 5 years
at the Technical University of Berlin, Germany as Research Fellow. Among the courses
taught at under graduate level include: Applied Mechanics, Mathematical Methods, Control
Engineering, Modeling and simulation of dynamic system, Smart Materials Reliability
Engineering as well as nanomechanics and nanomaterials. Professor Nikabdullah has
authored 4 books and 2 chapters in books as well over than 50 journal papers. He also
filed a patent at the Malaysian patent office under the title Anti-Fire Locking System 2002
Malaysian Patent, Status: Malaysian Patent Application Pending: Application Number
PI20022811. Date of Filing: 24th July 2002.
Contact: enikkeister.1@gmail.com
Professor Abdullah is the Head of Mechanical and Materials Engineering Department,

Universiti Kebangsaan Malaysia (UKM), Malaysia. He obtained the PhD degree from The
University of Sheffield in 2005, and his thesis is associated to the scope of fatigue data
processing and life assessment using the wavelet transform. The main research area of
Professor Abdullah is related to the structural integrity of metallic materials and automotive
components, specifically in the field of fatigue life assessment. He authored more than 200
journals/proceedings papers (indexed by ISI/SCOPUS) in the field of fatigue failure and
fatigue data analysis.
Contact: shahrum1@gmail.com
Professor Ariffin is an academic staff at the Department of Mechanical and Materials

Engineering, UKM. In 1995, he obtained his PhD from University of Wales Swansea
under the Mechanical Engineering Department and Institute of Numerical Methods in
Engineering. Professor Ariffin teaches Mechanics of Materials, Computational Methods
in Engineering and Finite Element Methods. His specialty is in computational method in
engineering under the area of powder mechanics, fracture mechanics, friction, corrosion,
finite element/discrete element and parallels computations. He is a Fellow of the Institute
of Materials Malaysia, founder of Malaysian Association of Computational Mechanics and
also member of International Association of Computational Mechanics (IACM).
Contact: kamal@eng.ukm.my
Mechanical Tolerance Synthesis on the Multimedia Oriented Systems Transport

Standard Transceiver Module, p.75-p.98
Dr. Michael Vervaeke, Els Moens, Dr. Youri Meuret, Prof. Heidi Ottevaere, Carl Van
Buggenhout, Piet De Pauw, Prof. Hugo Thienpont
Michael Vervaeke was born in Asse, Belgium, in 1975. He graduated as Industrial

Electromechanical Engineer from the Katholieke Hogeschool Sint-Lieven (Gent, Belgium)
in 1997. He enrolled subsequently for the Masters in Engineering at the Vrije Universiteit
Brussel (Brussels, Belgium) where he obtained his degree in Electrotechnical Engineer with
majors in Photonics in July 2000. That same year he joined the Applied Physics and Photonics
department of the VUB where he pursues research on the packaging and integration of
micro-optical and electrical devices. He developed a three-dimensional mechanical model
of an intra-chip optical interconnect module and is building a demonstrator platform using
in-house rapid prototyping technology (Deep Proton Writing). In 2007 he received a Ph.D.
on Photonic Modelling and Packaging of Chip-Level Interconnects) He authored more
than 56 publications in international conference proceedings in his research field as well as
in education, and authored or co-authored 11 journal publications. As Senior Researcher he
is involved in the public procurement and infrastructure buildup for new heavy equipment,
including an advanced ultraprecision 5-axis diamond tooling lathe and milling workstation
and a 300 mm hot embossing machine. He is B-Phots technical project leader for the
construction of a new cleanroom and laboratory environment..
Affiliation: Vrije Universiteit Brussel, Dept. Applied Physics and Photonics (TONA),
Brussels Photonics Team (B-PHOT), Pleinlaan 2, 1050 Brussel, Belgium
Contact: michael.vervaeke@b-phot.org
Els Moens was born in Vilvoorde in 1983. In 2006 she graduated as an Electrotechnical
Engineer with majors in Photonics. Her master thesis covered the design, fabrication and
characterization of micro-opto-mechanical components to improve the detection efficiency
in chromatographic systems. Since Octobre 2006, she is working towards a PhD at the
Department of Applied Physics and Photonics (TONA) of the Vrije Universiteit Brussel.
Her research is currently focused on miniaturized multichannel imaging systems with a
wide field of view. Els Moens is now working at ADB Airfield Solutions.
Brussels Photonics Team (B-PHOT), Pleinlaan 2, 1050 Brussel, Belgium / ADB Airfield
Solutions, Leuvensesteenweg 585, 1930 Zaventem, Belgium.
Contact: els.moens@adb-air.com
Youri Meuret graduated in Physics at Ghent University, Belgium, in 1998. In 2004, he

received his Ph.D. degree in Engineering from the same university, investigating the
contrast and color properties of digital projection systems in collaboration with BARCO
N.V. He then joined the Brussels Photonics Team (B-PHOT) at the Vrije Universiteit Brussel
(VUB), Belgium, where he is responsible for a research group which focuses on novel
display configurations and illumination systems. His current research interests include
3-D visualization and LED/laser based projection displays. In addition he explores the
potential of micro-optical components for novel solar concentrators and imaging systems.
Dr. Meuret is author or co-author of 20 SCI rated journal articles and 40 publications at
international conferences.
Contact: youri.meuret@b-phot.org
Heidi Ottevaere obtained in 2003 the PhD degree in applied sciences from the Vrije
Universiteit Brussel (VUB), Brussels, Belgium for her work entitled: Refractive
microlenses and micro-optical structures for multi-parameter sensing: a touch of micro-
photonics. Afterwards she continued her career as a post-doc researcher and from
2009 onwards as full professor at VUB. She made research efforts in a large variety of
fundamental and applied research topics, most of them situated in the domain of micro-
photonics and micro-optics. In particular she has built up expertise in the following research
topics: plastic micro-optics based on Deep Proton Writing, interferometric measurement
techniques, microlens characterization and biophotonics.
Contact: Heidi.Ottevaere@vub.ac.be
Carl Van Buggenhout was born in Aalst, Belgium, in 1983. He graduated as an

electrotechnical engineer in 2006 at the Universiteit Gent, Ghent, Belgium. Since 2006,
Carl Van Buggenhout is working for Melexis, a company active in the semiconductor
industry, as a development engineer. In this position he has been working on the design and
development of state-of-the-art IC packages for sensors, mainly in the optics domain and
since 2011, he is also focusing on wafer-level packaging of sensors. In 2012, he achieved
a PMP-certification.
Affiliation: Melexis Ieper N.V., Rozendaalstraat 12, B-8000 Ieper, Belgium.
Contact: cva@melexis.com
Piet De Pauw. Born in Dendermonde (Belgium) in 1956. He graduated as electronic

engineer in 1979 and as Ph.D. in 1984 from the University of Leuven. Subject of Ph.D.
thesis: Low Cost Solar Cells. He obtained an MBA from the Handels HogeSchool (now
Flanders Business School) in Antwerpen in 1990. After his Ph.D. research, he started
as a design engineer and project leader of high voltage circuits in Mietec (Oudenaarde,
Belgium) in 1984. In 1986 he joined Alcatel Bell (Antwerpen, Belgium) as head of the
reliability laboratory, failure analysis laboratory, IC technology, and Product Engineering.
In 1991 he joined Alcatel Microelectronics in Oudenaarde, Belgium, (now part of ON
Semiconductor) as Reliability Manager, and later Quality and Reliability Manager. Between
1998 and 2002 he established and expanded the company Sipex Flanders Design Center
in Zaventem Belgium, where he introduced new product lines. Starting from 2002 he is
a consultant specialized in optoelectronic product development (image sensors, infrared
sensors, Optical storage products, Opto datacom products). From 2000 till 2009 he was
a member of the board of directors of Q-Star Test N.V. in Brugge. From April 2006 till
2009 he chaired a division of Melexis N.V., and from 2006 till 2011, he was site manager
of Melexis Ieper.
Affiliation: Melexis Ieper N.V., Rozendaalstraat 12, B-8000 Ieper, Belgium.
Contact: pdp@melexis.com
Hugo Thienpont was born in Ninove, Belgium, in 1961. He graduated as an electrotechnical

engineer in 1984 and received the Ph.D. degree in applied sciences in 1990, both at the
Vrije Universiteit Brussel (VUB), Brussels, Belgium. In 1994 he became Professor at the
Faculty of Engineering. In 2004 he was elected Chair of the Department of Applied Physics
and Photonics. Currently he coordinates several research and networking projects at the
European level and manages micro-photonics related industrial projects in collaboration
with companies like Barco, Melexis, Best, Tyco, and Umicore. He authored more than
220 SCI-stated journal papers and more than 400 publications in international conference
proceedings. Prof. Thienpont received the International Commission for Optics Prize ICO
in 1999 and the Ernst Abbe medal from Carl Zeiss. In 2003, he was awarded the title of
IEEE Photonics Distinguished Lecturer. In 2005 he received the SPIE Presidents Award
for dedicated services to the European Photonics Community, and in 2007 the international
MOC Award for his contributions in the field of micro- optics and the prize Prof. Roger
van Geen for his scientific achievements during his research career at VUB. In 2008 he
obtains the prestigious Methusalem status for top-scientist from the Flemish government
for his research track-record in photonics. Hugo Thienpont is a member of the Board of
Stakeholders of the Technology Platform Photonics21, a high-level advisory board for
optics and photonics in Europe.
Contact: Hugo.Thienpont@vub.ac.be
Modeling Window Glass Automotive Antennas with EM Solvers. Experimental
Validation Using a Full-Sized Car Model, p.99-p.111
Dr. Matteo Cerretelli, Dr. Paolo Facchini, Dr. Andrea Notari, Prof. Guido Biffi Gentili
Dr. Matteo Cerretelli was born in Florence, Italy, in 1973. He received the Laurea degree
in electronic engineering from the University of Florence, Italy, in 1999 with a degree
thesis on the optimization and the synthesis of microwave structures employing neural
networks. Since 2000 he is member of the local Engineer Association. In 2000, he joined
the Department of Electronic and Telecommunications, University of Florence, Italy, where
he works on automotive antenna for ASK Industries S.p.A. His main research interests
are in the analysis and design of automotive antennas, microwave sensors and biological
applications of microwaves.
Affiliations:
- University of Florence, via Santa Marta 3 - I-50134 Firenze, Italy.
- ASK Industries spa, via P. Togliatti 46/A - I-42020 Montecavolo di Quattro Castella,
Reggio Emilia, Italy.
Contact: matteo.cerretelli@unifi.it
Dr. Paolo Facchini was born in Siena, Italy, on 5 Feb 1979, received the Laurea degree in
Telecommunication Engineering in July 2005 from University of Siena with a degree
thesis about Analysis and Design of Frequency Selective Surfaces. Starting 2006, focused
on Automotive antenna design collaborating with the most relevant automotive companies
in Italy. Currently works in R&D department of ASK Industries S.p.a.
Affiliation:
Contact: facchinip@askgroup.it
Dr. Andrea Notari was born in Scandiano, Italy, in 1974. He received the Laurea Degree
(cum Laude) in Electronic Engineering from the University of Parma, Italy, in 2001 with a
degree thesis on Optical Integrated Circuits. Since 2001, he has been with ASK Industries
S.p.A. working in Automotive Antenna R&D Department first as RF and Microwave
circuit designer and then, starting from 2006, as electronic department manager. Since
2006, he has been with IEEE association. His main research interests are in the analysis and
design of RF and Microwave antenna and circuit, in particular for automotive applications.
Affiliation:
Contact: notaria@askgroup.it
Prof. Guido Biffi Gentili was born in Lucca, Italy, on August 9th, 1943. He received the
Ph.D. degree in Electronic Engineering from the University of Pisa, Italy, in 1970. During
his graduate work at the University of Pisa, he held a research assistantship in the fields of
radar system modelling, active pulse compression circuits and biological applications of
microwaves. In 1976 he joined the School of Engineering, University of Florence, Italy,
where he is currently full Professor of Electromagnetic Theory and Techniques and Chair
of the Electronic and Telecommunication Department. His recent research activities have
focused on the modelling and design of active and passive multilayer planar antennas
and near-field sensors, wireless communication systems, millimetre wave imaging
techniques, industrial applications of high frequency high power EM fields, microwave
and RF hyperthermia and metamaterials EM modelling. He has authored over 100 refereed
journal and conference proceedings papers, reports and book chapters. Dr. Biffi Gentili
has been scientific consultant to many Italian companies, working in the fields of Defence,
Consumer and Medical Electronics and has authored 12 national and international patents
on vehicular compact antennas, RFID technologies and microwave sensors. Currently he is
responsible of many Research Projects committed by public and private companies in the
fields of microwave engineering.
Affiliation:
University of Florence, via Santa Marta 3 - I-50134 Firenze, Italy.
Contact: guido.biffigentili@unifi.it
Section 2 Passenger and Transportation Safety and Services
Spread Spectrum Radar Technology for Automotive Applications: State of the Art
and New Proposals, p.113-p.147
Dr. Susanna Spinsante, Prof. Ennio Gambi
Dr. Susanna Spinsante received her Ph.D. in Electronic Engineering and

Telecommunications in 2005 from Marche Polytechnic University (Ancona, Italy), and her
Laurea Degree in Electronic Engineering (summa cum laude) from the same University,
in 2002. Since 2005 she joined the Department of Electronics and Automation (now
Department of Information Engineering), as a Post Doc researcher, dealing with H.264/
MPEG4 video coding and compression applications, and security issues related to space
communications (authentication and encryption of TC/TM data). In the last two years,
her research activities focused on the fields of video techniques in AAL applications, and
innovative sets of sequences for radar and SS applications. During her Ph.D., she spent
six months at Department of Informatics of the University of Bergen (NORWAY), in the
framework of the FP6 European project FASTSEC Marie Curie Training Site on coding
theory and cryptography (contract n HPMT-GH-01-00260-10), on iterative coding
issues for asymmetric channels. In December 2004 she was one of the founders of the
academic spin-off SME, ArieLAB Srl; since January 2010 she is the CEO of the same
company. ArieLAB is involved in R&D activities in the field of home automation systems
and interfaces design, with specific focus on the integration of sensors and biomedical
devices, and multimedia applications. Susanna has been adjunct professor of the course
Telecommunication Services in the Academic Years 2007-2009, and teaching assistant
from 2003 to 2012, either at the Marche Polytechnic University and the University of
Trento. She collaborates as a reviewer for the IEEE Communication Letters, Journal of
Communication Systems and Software, and several other international journals. She was
and currently is involved in the Technical Program Committees of several international
conferences. She is co-author of more than 70 papers published in international journals,
and presented at international conferences.
Department of Information Engineering (DII)
Marche Polytechnic University, Ancona, ITALY
Via Brecce Bianche 12, I-60131, Ancona
Email: s.spinsante@univpm.it
Tel. +39 071 2204894
Prof. Ennio Gambi was born in Loreto (Ancona, Italy) in 1961. He received the Electronic
Engineering graduate diploma at the Universit Politecnica delle Marche in 1986 and
the Microwave Engineering master degree in 1989. From 1984 to 1992 he worked with
the Azienda di Stato per i Servizi Telefonici, while during 1988 he covered the role of
Avionic Engineer as an Official of the Italian Military Air Force. Since 1992 he joined
the Universit Politecnica delle Marche in Ancona, where he is, currently, an Associate
Professor in Telecommunications. At present his main research interests are in spread
spectrum systems, both for communication and radar purposes. In particular, his research
is dedicated to the selection of suitable sets of expansion sequences, able to increase the
system performance. At the same time, on the basis of a long time interest on source
coding, he is working on the evolution of this technique in the field of Ambient Assisted
Living, with the aim to propose an integrated solution where the fusion of data provided by
different sensor networks allows to offer the desired level of home assistance for disabled
or elderly people. He is coauthor of several international papers. Since the Academic Year
1995/96, he is the lecturer for the courses of Telecommunication Systems of the Universit
Politecnica delle Marche. He is a founding member of ArieLAB Srl, an Italian SME,
spinoff of the Universit Politecnica delle Marche, principally devoted to the development
of innovative technologies. In particular, ArieLAB products are in the field of home
automation solutions for disabled people and new systems for digital television.
Department of Information Engineering (DII)
Marche Polytechnic University, Ancona, ITALY
Via Brecce Bianche 12, I-60131, Ancona
Email: e.gambi@univpm.it
Tel. +39 071 2204845
Monitoring and Controlling Product and Service Engineering: The role of the Vehicles
Repair Solutions, p.149-p.161
Professor Nouha Taifi
Prof. Nouha Taifi is Assistant Professor at the Industrial Engineering Department,

Mohammadia School of Engineering, University Mohammed V, Rabat, Morocco where
she teaches courses on Innovation, Creativity and Change Management, and Process and
Business optimization and re-Engineering. She holds a PhD in eBusiness Management,
eBusiness Management Section Scuola Superiore ISUFI, University of Salento, Italy,
a MSc in Business and Economics, International University of Dalarna, Sweden, and a
BBA, University of AlAkhawayn University, Ifrane. She was a researcher at the Center
for Business Innovation, Universit del Salento, Lecce, Italy with a focus on the strategic,
technological and organizational aspects of the complex environments, and a postdoc
researcher at cPDM-laboratory-eBMS- Universit del Salento with a focus on the new
product development process in the automotive and aerospace industries. Her main research
interests are organizational structures and networks creation and management, information
and communication technologies engineering and management, and competences
development and management.
Affiliation:
- Nouha Taifi, Industrial Engineering Department, Mohammadia School of Engineering,
University Mohammed V, 10090, Rabat, Morocco.
Email: taifi@emi.ac.ma
VHDL-AMS Simulation of an AHRS Platform for Automotive, p.163-p.186
Mariagrazia Graziano, Massimo Ruo Roch
Dr. Mariagrazia Graziano

Dip. di Elettronica e Telecomunicazioni
Politecnico di Torino, Torino
Section 3 Electrical Engines and Environment Protection
Biolubricants and Triboreactive Materials for Automotive Applications, p.189-p.205

Amaia Igartua, Xana Fdez-Prez, Olatz Areitioaurtena, Rolf Luther, Christian Seyfert,
Jrgen Rausch, Iaki Illarramendi, Michael Berg, Hubert Schultheib, Benjamin Duffau,
Sylvain Plouseau, Mathias Woydt
aigartua@tekniker.es
Fundacin TEKNIKER
Avda. Otaola, 20
20600 Eibar
Guipuzcoa
Spain
Design Methodology for Small Brush and Brushless DC Motors, p.207-p.235

Jrme Cros, Mehdi Taghizadeh Kakhki, Geraldo C.R. Sincero, Carlos A. Martins, Philippe
Viarouge
The authors are all with Laval university and are members of LEEPCI (Laboratoire
dlectrotechnique, dlectronique de Puissance et de Commande Industrielle).
Prof. Jrme Cros is the corresponding author and could be contacted at: LEEPCI,
Electrical & Computer Eng. Dept., Laval University, 1065 avenue de la Mdecine, Quebec,
G1V 0A6, Canada;
Email: jerome.cros@gel.ulaval.ca.
Development of a Co-Sensor for Hydrogen Fuel Cell Powered Vehicles, p.237-p.250

G. Neri, G. Leonardi, M. Parthibavarman, V. Hariharan, C. Sekar
G. Neri received the degree in Chemistry in 1980. He spent several periods at the University
of Michigan (USA), and in 2001 has been appointed Full Professor of Chemistry. He has
been Director of the Department of Industrial Chemistry and Materials Engineering of
the University of Messina. His research activity covers many aspects of the synthesis,
characterization and chemical-physics of materials with particular emphasis to catalytic
and sensing properties. In the latter research area his work has been focused on the
preparation of metal oxide thick and thin films, organic-inorganic hybrid nanocomposites,
novel coordination complexes and their application in gas sensors in collaboration with
several research groups in Italy and abroad.
Dept. of Industrial Chemistry & Materials Engineering, University of Messina, 98166
Messina, Italy.
E-mail: gneri@unime.it
S. G. Leonardi received the degree in Materials Engineering from the University of Messina
in 2011. He is currently a Ph.D. student of Engineering and Chemistry of Materials at
University of Messina. His research activity is focused on the development of chemical and
biochemical sensors based on nanostructured materials.
Dept. of Industrial Chemistry & Materials Engineering, University of Messina, 98166
Messina, Italy.
E-mail:leonardis@unime.it
M. Parthibavarman is currently working as an Assistant Professor in Mahendra Engineering

College, Kallipatti, Salem. He received his M.Sc. and M.Phil. degrees from Bharathiar
University, Coimbatore through Sri Ramakrishna Mission Vidyalaya College of Arts and
Science. He is likely to submit his doctoral thesis to Periyar University soon. His research
field includes the synthesis of metal oxide based semiconducting nanoparticles for sensor
applications.
Centre for Nanoscience & Technology, Department of Physics, Periyar University,
Salem-636011, India.
E-mail: varmanphysics85@gmail.com
V. Hariharan is pursuing his doctoral programme in the Department of Physics, Periyar

University, Salem. He has been selected as Project Fellow in the University Grants
Commission (UGC) sponsored Non-Recurring Special Assistance Program. His research
fields include metal oxide based semiconducting nanoparticles for biosensor and
supercapacitance applications.
Centre for Nanoscience & Technology, Department of Physics, Periyar University,
Salem-636011, India.
C. Sekar received his Ph.D from Anna University, Chennai in the year 1997. Subsequently
he has spent several years in the R & D laboratories of NTT Corporation, Japan and at the
Institute for Solid State and Materials Research (IFWDresden), Germany. He is interested
in new materials research. He has worked on several novel materials such as fullerences,
CNTs, biomaterials, spin ladders, sodium cobalt oxide, semiconducting metal oxides,
etc. His recent research activity includes synthesis of nanostructures and fabrication of
chemical and biosensors.
Department of Bioelectronics and Biosensors, Alagappa University, Karaikudi-630003,
India.
E-mail: sekar2025@gmail.com
Table of Contents
p. 17 Preface
Section 1
New Technologies and Advances in Automotive Engineering
p. 21 The Current Status of Automotive Coatings and Future Trends Using

Nanotechnology
p. 53 Automotive Lower Suspension Arm: Experimental and Simulation

Durability Assessment
N. A. Kadhim, N. Nikabdullah, S. Abdullah, A.K. Ariffin
p. 75 Mechanical Tolerance Synthesis on the Multimedia Oriented Systems

Transport Standard Transceiver Module
Dr. Michael Vervaeke, Els Moens, Dr. Youri Meuret, Prof. Heidi Ottevaere,
Carl Van Buggenhout, Piet De Pauw, Prof. Hugo Thienpont
p. 99 Modeling Window Glass Automotive Antennas with EM Solvers.

Experimental Validation Using a Full-Sized Car Model
Dr. Matteo Cerretelli, Dr. Paolo Facchini, Dr. Andrea Notari, Prof. Guido
Biffi Gentili
Section 2
Passenger and Transportation Safety and Services
p. 113 Spread Spectrum Radar Technology for Automotive Applications: State

of the Art and New Proposals
p. 149 Monitoring and Controlling Product and Service Engineering: The

Role of the Vehicles Repair Solutions
p. 163 VHDL-AMS Simulation of an AHRS Platform for Automotive

Section 3
Electrical Engines and Environment Protection
p. 189 Biolubricants and Triboreactive Materials for Automotive Applications

Amaia Igartua, Xana Fdez-Prez, Olatz Areitioaurtena, Rolf Luther,
Christian Seyfert, Jrgen Rausch, Iaki Illarramendi, Michael Berg, Hubert
Schultheib, Benjamin Duffau, Sylvain Plouseau, Mathias Woydt
p. 207 Design Methodology for Small Brush and Brushless DC Motors
Jrme Cros, Mehdi Taghizadeh Kakhki, Geraldo C.R. Sincero, Carlos A.
Martins, Philippe Viarouge
p. 237 Development of a Co Sensor for Hydrogen Fuel Cell Powered Vehicles

Prof. Giovanni Neri, G. Leonardi, M. Parthibavarman, V. Hariharan, C. Sekar
Preface
How to start writing and defining the role of vehicles in world today? Is it better to mention
numbers, statistics, facts, reports, data.perhaps fuel emissions should be the hottest first
sentence as an eye catcher, or new technologies and advances in the field? Whatever one
chooses, there is so much to say; far more than a preface can offer, or that a preface should
contain.
It is estimated that for the first time in history, this year, over 60 million passenger cars will
be produced making up over 74% of total annual vehicle production. This means over 60
million people will own a new vehicle making their travels safer with better fuel efficiency
than ever. We often read car magazines, watch racing on TV and show off with our new
engine underneath the hood. Those are the moments when enormous number of questions
pop out, thinking what is next? Where is the limit? Are the producers up to the task; or
can they improve? Nevertheless, is the fuel efficiency making a difference in home budget
taking in consideration sky rocketing oil prices? Perhaps we are witnessing the dawn of
vehicles powered by new energy sources?
Indeed, many excellent questions. Some of the answers (or possible answers) can be
found here, within covers of this book, divided into three sections: New Technologies and
Advances in Automotive Engineering, Passenger and Transportation Safety and Services,
and Electrical Engines and Environment Protection.
Each section covers novelties, discussions and engineering solutions to specific problems
and issues. Authors, each and every one a master in the field, have tried (and did a great
job) bringing their chapters both to a wider audience as well as to the specific niche of
researchers, engineers and other experts making this book a great addition to many shelves.
Every single chapter is a joy to read, and I am sure many readers will cite them more than
once.
AcademyPublish.org - Vehicle Engineering 17

Section 1
New Technologies and Advances in Automotive Engineering
The Current Status of Automotive Coatings and Future Trends Using

Nanotechnology
Automotive Lower Suspension Arm: Experimental and Simulation Durability

Assessment
N. A. Kadhim, N. Nikabdullah, S. Abdullah, A.K. Ariffin
Mechanical Tolerance Synthesis on the Multimedia Oriented Systems

Transport Standard Transceiver Module
Dr. Michael Vervaeke, Els Moens, Dr. Youri Meuret, Prof. Heidi Ottevaere, Carl Van
Buggenhout, Piet De Pauw, Prof. Hugo Thienpont
Modeling Window Glass Automotive Antennas with EM Solvers. Experimental

Validation Using a Full-Sized Car Model
Dr. Matteo Cerretelli, Dr. Paolo Facchini, Dr. Andrea Notari, Prof. Guido Biffi Gentili
The Current Status of Automotive Coatings and Future
Trends Using Nanotechnology
M.Mohseni, H. Yari, B. Ramezanzadeh
ABSTRACT
Today, automotive coatings are considered as an important market and play a major role
in car industries. They not only impart aesthetic and a pleasant appeal to car bodies, also
protect them against degrading environmental effects such as acid rain, hot-cold shocks,
UV radiation, stone chips, car washing, fingernails, blowing sands, keys and branches.
Their use extends from metallic parts to plastic vehicle components both as exterior and
interior types. This chapter aims at reviewing the status of coatings in this sector in recent
years and intends to report the advances achieved. Future trends utilizing nanotechnology
will also be reported. Examples include scratch/wear resistant clear coats, non-hazardous
chromium-free pretreatment coatings, self-healing repair finishes, abrasion resistant
coatings for head lamps, soft feel coatings for interior plastic front panels, self-cleaning
top coats, water repellent, heat and anti-reflective and anti-fog coatings for windscreen
and mirrors, electrochromic and smart coatings for windows, scratch proof coatings for
plastic car wheels, UV curable and powder coating systems, as well as iridescent and
color changing base coats. For all these applications emphasis will be made on the
physico-chemical, mechanical and weathering properties of these coatings.
AN INTRODUCTION TO AUTOMOTIVE COATINGS
Two main goals are envisaged when coatings are applied to substrates. The first involves
protecting the substrate against various aggressive environmental agents (such as
sunlight, humidity and corrosive materials) and the second includes imparting color and
aesthetic appeal to the substrate. In some applications, these two are highly important
such as in automotive coatings. Extensive exposure to different permanent environmental
factors during service life as well as occasional (acid rains and various biological
substances) makes the first goal highly prominent in automotive coating industries. In
addition, a pleasant appeal is vitally important for coatings in automotive industries to be
acceptable by customers.
Coating materials have been developed progressively in recent years. The earliest
automobile coatings were those used on furniture and horse drawn carriages. Their
applications were labor intensive and their color variations and the durability were
limited. According to the different advances started from the earliest systems up to now,
the time line for automotive coatings includes varnishes, nitrocellulose, alkyd resins,
lacquers, acrylic resins, electrocoats (e-coats), primers, topcoats (base-coats plus clear-
coats), smart coatings and nano-based coatings as well as green technologies.
Automotive coatings are considered a multi-layer system in which each layer has its own
especial function. These make the whole system resist to various environmental factors

mentioned above. Fig. 1 shows a typical automotive coating system which is currently
applied in most of automotive paint plants.
Figure 1. Specifications of a multilayer automotive system
As Fig. 1 describes, the substrate is firstly coated by a conversion phosphating or

chromating coating to enhance the adhesion and corrosion protection of the substrate.
Then, an electro deposition (ED) coating, usually based on epoxy-amine containing
anticorrosive pigments is applied. The primer layer which is a polyester melamine
coating is then applied which its main function is enhancing the coating system resistance
against mechanical deformations such as stone chipping. The color and special effects,
such as metallic luster are imparted to the automotive coating by basecoat layer which is
typically an acrylic melamine resin. The final layer of an automotive coating system that
plays an important role in both aesthetic and protective aspects is the clearcoat layer. It is
usually a clear acrylic/melamine or polyurethane coating. As there is no pigment or
micron-size fillers in this layer a highly smooth and glossy surface is obtained. On the
other hand, the clearcoat apart from creating a highly glossy surface, is intended to
protect the underneath layers and even the substrate against various aggressive
weathering (i.e. humidity and sunlight) and mechanical (i.e. mar and scratch) factors
during service life.
Current and future trends in automotive coatings can be generally categorized in two
main classes; environment-related and functional-related issues. The former deals with
the developments in formulations that are benign or emit less or no volatile organic
compounds (VOC) to the environment and the latter refers to the inventions in which the
coating functions a new duty rather than the protection and beauty. In this chapter, these
advances are shortly reviewed.
ENVIRONMENTAL-RELATED ISSUES
The advances related to environment are explained into two main categories; eco-friendly
conversion coatings and eco-friendly organic coatings.
Eco- Friendly Conversion Coatings

Organic coatings have been widely used to protect metals in the manufacturing and
construction industries from corrosion. It has been realized that surface pretreatment of
the metal substrates seems to be required before paint application in order to promote
paint adhesion and improve corrosion performance. Different kinds of conversion
coatings have been used in recent years. Various factors including characteristic
performance, price, application process and environmental issues must be considered
22 AcademyPublish.org - Vehicle Engineering

when a conversion coating is used (Li, 2010; Marsh, 2001; Harun, 2005; Bajat, 2007;
Bajat, 2008; Srensen, 2009; Van Wazer, 1967; Saravanan, 2009; Cho, 2009; Narayanan,
2005).
Conversion coatings like chromate and phosphate (Figure 2) have been extensively used
for many years to protect car bodies from corrosion and improved the automotive
coatings adhesion to the metal substrates (Li, 2010; Marsh, 2001; Harun, 2005; Bajat,
2007; Bajat, 2008).
The adhesion of an automotive coating to the metal substrate depends on both the
chemical and physical characteristics of the coating and substrate (Li, 2010; Marsh, 2001;
Harun, 2005; Bajat, 2007). Automotive coatings, due to their good wetting properties,
can produce excellent physical adhesion. The higher adhesion of an automotive coating
can be achieved when the coating reacts with the metal. One possible way to improve the
physical as well as the chemical adhesion forces between the metal and the coating is to
use chemical treatments. Using conversion coating (CC), the surface of the metal can be
cleaned from different kinds of contaminations which are responsible for the decrease in
the paint film adhesion. The surface tension and therefore wetting properties of the metal
can be also significantly affected in presence of conversion coatings (Li, 2010; Marsh,
2001; Harun, 2005; Bajat, 2007; Bajat, 2008; Srensen, 2009; Van Wazer, 1967).
Figure 2: Diagrammatic representation of phosphating reaction (left) and scanning

electron microscope (SEM) micrograph of the phosphate coating morphology (right) (Li,
2010; Marsh, 2001).
The chromate based conversion coatings have been widely used for many years. Among
them Cr(VI) pre-treatment due to its superior corrosion resistance, good paint adhesion
and low cost (Cho, 2009; Narayanan, 2005) has been used extensively. This kind of
chemical treatment contains soluble Cr+6 spices which is responsible for the self-healing
behavior. However, the Cr(VI) coating consists of carcinogenic materials. Therefore,
because of the high toxicity of this kind of CC, attempts have been made to find other
more environmentally friendly alternative pretreatments (Cho, 2009; Narayanan, 2005).
The phosphating pretreatment has been used as most common surface pretreatment for
both ferrous and non-ferrous metals. It has been shown that this pretreatment plays a vital
part in the automobile, appliance and general manufacturing industries. However, there
are some limitations during phosphating process (Van Wazer, 1967; Saravanan, 2009).
Concentrated phosphate bath has a detrimental impact on ground and surface water
ecology and the sustainability of water resources. While costly waste treatment and
disposal are in use, phosphate-free environmentally-friendly solutions are also desired.

Significant sludge generation in the concentrated phosphate bath necessitates frequent
sludging (Van Wazer, 1967; Saravanan, 2009; Cho, 2009). Phosphate conversion coating
is usually porous and requires an additional sealing procedure. Typically phosphating is
carried out at higher temperature which requires expensive energy.
New Pretreatments
Because of environmental regulations in recent years, research on new surface
pretreatments (with lower VOC) to replace chromate and phosphate conversion coatings
has been performed.
One promising new pretreatment is the application of zirconium oxide (ZrO2) on metal
surface by the sol-gel method or rinsing with hexafluorozirconic acid (H2ZrF6) solution
(Hirai, 1998; Di Maggio, 2001; Fedrizzi, 2001; Li, 2001; Gusmano, 2007). It has been
shown that sol-gel zirconia coatings could provide similar or even better corrosion
protection performance compared with the phosphate coatings by promoting adhesion.
The Cr(III) pretreatment has been another opportunity to be used instead of Cr(VI). It is
well known that Cr(III) pretreatment unlike Cr(VI) is less toxic and carcinogenic.
However, it has been found that the corrosion resistance of the Cr(III) treatment may not
be as good as the Cr(VI) (Saravanan, 2009; Cho, 2009; Ramezanzadeh, 2010a;
Ramezanzadeh, 2011b). Attempts have been made to improve the anticorrosion
performance of the Cr(III) treatment. Font, et al. used transition metal ions such as Co(II),
Ni(II) and Fe(II) in Cr(III) bath to improve the Cr(III) coating anticorrosive properties
(Ramezanzadeh, 2010a; Ramezanzadeh, 2011b).
Automotive Electrocoat Based Coatings

Electrodeposition coatings have been used on the pre-treated car bodies by applying an
electrical potential to an aqueous dispersion of charged particles since the mid 1970s
(Corrigan, 1992). The electrodeposition coating provides good adhesion of organic
automotive primer over metal car body. This kind of coating, which is based on
waterborne cationic epoxy electrocoats, could significantly improve corrosion resistance
and produce less environmental contaminations. Low coverage of electrocoatings on
sharp edges, during baking process, of a car body is the most important problem resulted
by surface tension of this kind of coating. Researchers have utilized many techniques to
overcome this problem: (a) restricting the flow characteristics of the resin, (b) using
double coating techniques with separate baths and (c) using double layers which are
deposited one after the other from the same bath. Restricting the flow of electrocoating
may lead to greater car body edges coverage. However, the smoothness will be reduced
by lowering coating flow. This problem can be eliminated by having a two layer system
(Corrigan, 1992; McMillan, 2002; Berger, 1984). Applying second electrocoating layer
(with good flow) in the second bath provides appropriate smoothness at edges. Decrease
in electrocoating coverage at sharp edges results in decrease in corrosion resistance. It has
been shown that increase in pigment content (pigment/binder (p/b) ratio) of the
electrocoating results in decreased coating flow during curing stages, leading to better
film coverage over the sharp edges of the substrate. This results in improved corrosion
resistance at sharp edges. However, lowered coating flow using pigment leads to a more
textured and orange-peeled coating surface, affecting the final appearance. Therefore, to
obtain an appropriate coating smoothness, the pigment content could not be increased in
order to observe desired edge coverage. Voids and micro sized pinholes will be produced
using high pigment content in the electrocoating formulation. The holes and voids are the
conductive pathways for the corrosive electrolyte permeation through the coating,
especially at sharp edges having low coating coverage. It has been also found that the
difficulty of dispersing high levels of inorganic pigments in water may lead to shear

instability and increase in settling on horizontal surfaces (McMillan, 2002; Berger, 1984;
Lister, 1992; Roberts, 1964; Schmael, 1988).
In recent years, new types of cathodic electrocoatings have been developed to improve
appearance, mechanical properties and to reduce the environmental problems. One way
to overcome this problem is using microgels (Fig 3), which are high molecular weight
resins dispersed in water, in order to achieve appropriate flow restriction (comparable
with the case of using high p/b) whilst reaching an acceptable film appearance. Cationic
epoxy microgels can be produced by the reaction between oligomer with an amine.
Adding a counter acid to the water containing epoxy microgels results in an emulsion
creation.
Fig 3: Microgels observed by a TEM technique.
Autophoretic conversion coatings (ACC)

Autophoretic processing has been introduced as a waterborne process which can be used
instead of conventional electrocoating one in which the phosphate surface pre-treatment
would not be required prior to coating application. The bath of autophoretic system
includes distilled water, a slightly acidic latex emulsion polymer and other ingredients.
The solid content of the bath is about 3-5 % whilst the viscosity is similar to water. The
first step of bath solution reaction to the metal surface in autophoretic coating process is
iron oxidation due to the bath acidic condition. The released iron cations are able to react
with bath solution and be deposited on the metal substrate. This process is schematically
shown in Fig 4.
The deposited organic film can strongly adhere to the metal surface showing a porous
structure. The porous structure of the deposited film allows the chemical activators
presented in the bath to diffuse rapidly into the metal/film interface, leading to further
metal oxidation. The process will be continued until the metal oxidation process, after
which iron oxide creation can be continued. Increase in coating thickness, resulting in a
high degree of surface protection and corrosion resistance, can be obtained in this
process. Depending on the bath temperature and although the time of process, the
thickness of the final coating can vary from 0.6 to 1 mm. There are some advantages in
using this technique including the uniform film formation even in difficult-to-reach areas
of the metal substrate. Therefore, unlike electrocoating system, the coating can be
uniformly produced even at complex shaped parts of the metals. Controlling the type and
concentration of reactants in the autophoretic bath, the film thickness as well as film
morphology can be controlled. The final film produced in autophoretic system consists of
two film layers: the coating layer with a high cohesiveness and the one related to

undeposited polymer film including activators presented in the bath. The reaction in the
second layer will be continued until the increase in film thickness is obtained. There are
many advantages in using autophoretric system compared with electrocoating systems
which are listed below:
Fig 4: Schematic illustration of the different stages of film formation in an autophoretic

method.
(a) eliminating conversion coating layer like phosphate layer

(b) saving energy
(c) decrease in operating cost
(d) decrease in toxic materials
(e) obtaining uniform film formation at sharp edges
(f) film formation at the difficult-to-reach areas and also complex parts of the metal
From the above explanations, the autophoretic system can be an opportunity to be used
instead of electrocoatings or be developed for new applications.
Eco-friendly Organic Coatings

There is a major effort in car manufacturers to replace current coating methods to reduce
VOC emissions. Several car manufacturers have developed alternatives to classical
painting methods. The conventional coating systems were usually liquid and their solid
content was too low (as low as 30 %). In these coating formulations solvents were a
major part of ingredients released during application and curing, producing a harmful
environment for both local and global atmosphere. VOCs have been the focus of
environmental concern and regulated for two basic reasons: human health issues due to
some VOCs being toxic and ozone formation through photochemical reactions with NOx.
Most of industrial volatile organic emissions come from the car body paint shops in final
assembly plants or from cleaning processes. Car manufacturers have developed a number
of solutions to reduce these emissions to comply with current legislations. These are:
paints with a low solvent content, water based paints, IR- or UV- curing systems.
UV-curing coatings
A new technology which has drawn attention of researchers in both academia and
industries in recent years and is commercially in use now is Ultraviolet (UV) curable
coatings. UV-cured coating technology is one of the fastest-growing technologies in the
coatings industry (Tyler, 2010; Meine, 2007). These formulations typically contain an
acrylated oligomer based on a polyether, polyester or epoxy resins (Seubert, 2004;
Stropp, 2006; Maag, 2000)). These coatings harden only a few minutes after being
exposed to ultraviolet lamps.

UV-curable systems have outstanding characteristics which make them a powerful
technology to be used in different areas of organic coating industry especially for
automotive coatings. The most important and unique advantages of UV-curable coatings
are their fast curing, low-temperature processing, no pot life issues, reduced and solvent-
free formulations and their scratch and chemical resistance. Additionally, they do not
need the use of high energy-consuming drying ovens. All of these factors make UV
coatings environmentally friendly. UV-coated parts can be further processed, stacked or
packed immediately after the UV-curing process.
In spite of various merits of utilizing UV-curable systems, there were some difficulties
and drawbacks in earliest formulations which have been removed or modified in recent
years. Some of these problems were: (a) the inferior weatherability of initial oligomers
with polyether, polyester or epoxy backbone which limited some applications where
excellent UV stability is required, (b) inefficient or no curing at some parts of the car
body which are inaccessible for the UV rays to reach, and (c) poor adhesion and
flexibility due to shrinkage during the curing process (Beck, 2006).
In order to overcome these problems, different solutions were utilized. Dual curing
systems were one of these modification ways that combine the benefits of UV and 2K
urethane systems. These coatings typically compose of acrylated polyisocyanates and
unsaturated polyester polyols (Marinovic, 2010; Simi, 2008). The system is firstly cured
by UV rays. It provides an instantaneous, tack-free film that because of less time for
airborne impurities to be deposited, the coating appearance is improved. The chemical
cure is performed by a thermal process in which isocyanate and hydroxyl functional
groups react to each other. Due to lower shrinkage in such curing processes, the cured
films are more flexible and the adhesion to the substrate is better than a 100% UV-
curable systems. In addition, thermal curing provides efficient reactions between
functional groups in areas that are not fully exposed to UV rays. In several studies, novel
approaches have been utilized to replace the less weatherable oligomeres such as
polyether, polyester or epoxy structures with those of more weathering-resistant ones like
acrylate urethane systems (Schwarzentruber, 2002; Snuparek, 2002; Valet, 1999). Theses
advances make UV-curable system to be used in automotive topcoat applications.
Developing UV cure coating technologies for both automotive OEM and refinish markets
has been targeted and even a brighter future is seen. This technology enables automotive
manufacturers to paint or repair more cars in a much shorter time with the much
friendlier formulations.
PPG, NEXA Autocolor, Bayer and BASF Companies currently use UV curable
technologies for various parts of automotive car body such as primer-surfacer layer and
topcoat refinishes. More recently, Bayer demonstrated that UV curable coating systems
can also be used to finish the polycarbonate headlamp assemblies on most vehicles.
Powder coating
Powder coatings, which are called the coating of the future, are finely grounded paint
particles. They are basically applied by spraying over a pre-heated substrate. They were
developed in the 1950's. The main benefit of these coatings compared to traditional liquid
ones is the absence of solvent in the formulation which is one step forward in both
environmental and economic issues of coating system (Orzech, 2007). Elimination of
solvent leads to a significant decrease in flammability of the formulation. Moreover
personnel involved in production line are not exposed to solvent. In addition, due to
energy and labor cost savings, operating investments are lower. The cost of applying
powder coating is one tenth of that of liquid coating. These merits along with high
durability of powder coatings have turned these coating systems a promising alternative
for replacing the conventional coating systems. Developments in variety of different

chemistries have resulted in a rapid growth in different applications (Shutic, 1998). Five
main families of powder coatings are epoxies (corrosion resistant, excellent chemical and
mechanical properties, exceptional adhesion), polyesters (tough and durable), epoxy-
polyester hybrids (similar to epoxies with improved weatherability), polyurethanes
(outstanding appearance and toughness with excellent weathering resistance, superior
chip, mar and scuff resistance and highly resistant against humidity and salt spray) and
acrylics (outstanding weathering resistance). Powder coating is a powerful option in
applications where a high resistance to chemical and environmental degradation is
necessary.
It was early 1980s when powder coatings were firstly used in automotive coatings
(Schmidt,1996; Bell, 1998). The automotive industry has witnessed use of powder
coatings for economic, quality, and environmental reasons. Among different markets of
powder coatings, automotive application had the fastest growth during that period till the
start of 21 century. In 2005, automotive applications for powder have been nearly 25
percent of total powder coating demand. This was less than 16 percent in 1994. The
automobile parts which are being powder coated include roof racks, door handles, oil and
fuel filters, brake pads, engines, exterior and interior trim, wheels, hubcaps, radiators,
bumpers, shock absorbers, mirror frames, battery trays, and coil springs. These
developments will help to assure that powder coatings market will continue to grow at an
impressive pace. Powder is being used in the automotive industry for the exterior bond
intermediate coat known as "primer surfacer", as well as for finishing of underhood
components. For example, BASF Coatings utilized a powder slurry process to apply
primer to automobile bodies at a German Mercedes-Benz plant (Biller, 2009). PPG and
Siebert Powder Coatings have developed and supplied powder primers to Chrysler and
GM respectively. Meanwhile, DuPont and BASF collaborated with the European
automobile manufacturers (Barberich, 2004).
Clear powder coatings, as an alternative to solvent borne clear coats, are being used for
automotive applications such as finishing wheels. Clear coats are also being used over
automotive exterior basecoats by some high end manufacturers. PPG Company has been
the worlds leading in powder automotive clearcoat. In recent years, they manufactured a
wide range of special and metallic effect powder coatings using mica and aluminum
flakes.
Developments of new formulations which can be cured at lower temperatures such as IR
cured powders, has significantly extended the market to heat-sensitive substrates such as
wood, plastics, and assembled components. Clear, corrosion resistant, and durable
powder coatings are used for a wide range of applications including automotive parts.
Infrared (IR) and ultraviolet (UV) curing technologies allow production speed to
increase.
FUNCTIONAL-RELATED ISSUES
Natures mysteries have been the origin of lots of inventions in different fields especially
automotive coatings. There are some natural models for creating coatings which possess
weird characteristics such as self-cleaning, anti-reflecting attributes or improving some
properties such as drivers night vision. Most of these inventions have been successfully
mimicked for creating functional coatings. Nanotechnology has had a key role in creating
these functions.
Nanotechnology can be described as the science of molecular engineering. The word

Nano is originated from the Greek word "small". A nanometer is one- billionth of a

meter. For comparison purposes, a human hair is roughly 80,000 nanometers. The finer
the particle, the closer they reside together. This nano-structure alters the surface area and
material characteristics dramatically. Nanotechnology has induced revolutions in
different fields of science especially in surface science. The use of nano-coatings, is
highly interesting. Furthermore, nano-coatings are expected to bring new improved
surface quality and to add interesting features to the surfaces. Some of these coatings are:
Scratch resistant clearcoats, soft feel coatings, self cleaning coatings, anti-reflecting
coatings and anti-fog coatings.
Scratch/Wear Resistant Clearcoats
Car industry has converted monocoat technology almost exclusively to basecoat/clearcoat

system in the last two decades. Basecoat/clearcoat technology utilizes finishes that
combine superior metallic appearance with excellent surface smoothness and gloss.
However, the appearance and visual performance of the clearcoat tends to be spoiled
during service life when exposed to harsh environmental conditions. Therefore, the need
to produce coatings with high capability of retaining their initial appearance is vitally
important with respect to weathering condition (i.e. sun, oxygen and water) as well as to
chemical and physical damages they experience (Amerio, 2008; Bautista, 2011; Barletta,
2010; Courter, 1997; Ramezanzadeh, 2011c; Ramezanzadeh, 2011d; Ramezanzadeh,
2011e; Yari, 2009; Tahmassebi, 2010; Shen, 2004). Scratch and mar are the most
important types of mechanical damages which have caused serious challenges for the
coatings formulators (Ramezanzadeh, 2011c; Ramezanzadeh, 2011d; Ramezanzadeh,
2011e; Yari, 2009; Tahmassebi, 2010; Shen, 2004). These have become a critical issue
for todays car manufactures too. Mar refers (Fig 5) to the light surface damages (less
than 25 m) whilst scratch refers to the medium to severe damages (below 1mm). The
scratch/mar in the clearcoat will scatter visible light and are appeared bright, causing the
affected area to be easily visible to the naked eye (Ramezanzadeh, 2011c;
Ramezanzadeh, 2011d; Ramezanzadeh, 2011e; Yari, 2009; Tahmassebi, 2010; Shen,
2004; Schulz, 2003; Hara, 2000; Jardret, 2000; Weidian, 2001; Thorstenson, 1994;
Ramezanzadeh, 2010f).
Fig 5: Different types of mechanical damages including mar/scratch, fracture and plastic
damages (Bautista, 2011).

Mechanical damages such as scratches may be caused by polishing equipments, carwash
bristles, tree branches, car keys, etc. The mechanism of this failure during scratch
condition is schematically shown in Fig 6.
Fig 6: Schematic illustration of scratch and mar damages occurred on an automotive

clearcoat (Barletta, 2010).
Therefore, scratch resistance is considered as the most important criterion for ranking the
performances of basecoat/clearcoat automotive paint systems. The strategies to measure
and characterize scratch/mar types damages have been widely developed over the past
few years. Depending on the viscoelastic properties of the clearcoat as well as on the
scratching condition (the cause of scratch, scratches force, scratch velocity and
environmental temperature) scratch/mar can be produced by two primary mechanisms:
fracture and plastic flow (Fig 7). Fracture types scratches are irregular and of a fractured
nature and may involve material loss, while the latter is smooth, regular and involves
plastic deformation of clearcoats.
The visibility of fracture-type scratches depends on the direction of incident and
observational lighting. Conversely, plastic-type scratches are not visible if the
longitudinal direction of the scratch coincides with the direction of the observational
lighting. Furthermore, there is a possibility that plastic-type scratches may be repaired by
sunshine heating during daylight hours.
Fig 7: Plastic (a) and fracture (b) types scratches produced on an automotive clearcoat
(Jardret, 2000).

Methods Of Improving Clearcoats Scratch/Mar resistance
From the explanations in the previous part, working on the methods of improving scratch
resistance and controlling the scratch morphology are of utmost importance in the
research and development departments of the automotive finishing industry. Accordingly,
researchers have proposed various methods for improving the scratch resistance of
automotive clearcoats. The proposed methods include procedures to increase surface
slippage and hardness, as well as enhancing cohesive forces within clearcoats that solely
modify the viscoelastic properties of clearcoats. Changing the chemical structure of a
clearcoat would guarantee modification of its viscoelastic properties (Ramezanzadeh,
2011c; Ramezanzadeh, 2011d; Tahmassebi, 2010; Ramezanzadeh, 2011e). However,
such chemical changes are not easily achievable. Furthermore, changing the chemical
structure of a clearcoat may result in unwanted adverse effects on other properties of the
resultant clearcoat and will in most cases, increase its price. Consequently, attempts have
been made in many research programs to modify viscoelastic properties by physical
incorporation of various additives and pigments into a clearcoat (Ramezanzadeh, 2011c;
Ramezanzadeh, 2011d; Ramezanzadeh, 2011e). It has been shown that controlled use of
the additives/pigments could ensure minimization of unwanted variations in other
properties of the resultant clearcoat as well as being an attractive, economically viable
alternative. Depending on the chemical/physical characteristics of the used additives
and/or pigments, they may modify the surface and/or the bulk properties of a clearcoat.
Additives like polysiloxanes and polyacrylates have been produced to increase surface
slipping and surface hardness and consequently increase the force necessary for creating
scratches. These may alternatively modify the intrinsic viscoelastic properties of the
clearcoat (Ramezanzadeh, 2011c). Various factors i.e. chemical composition of the
additives, additives end and side groups and additives molecular weights have been found
effective to influence their migration ability to the bulk and/or surface of the clearcoat. It
has been also found that incorporation of inorganic pigments such as micro-SiO2, micro-
TiO2, micro-Al2o3 and micro ZnO into the coating formulation may also improve
coating scratch resistance. However, micro-sized pigments results in unwanted scattering
and reduce the appearance. Therefore, the use of these pigments is only limited to other
types of coatings (Amerio, 2008; Bautista, 2011; Barletta, 2010; Courter, 1997;
Ramezanzadeh, 2011c; Ramezanzadeh, 2011d; Ramezanzadeh, 2011e; Yari, 2009;
Tahmassebi, 2010; Shen, 2004; Schulz, 38; Hara, 2000; Jardret, 2000; Weidian, 2001;
Thorstenson, 1994; Ramezanzadeh, 2011e; Dhoke, 2009a; Dhoke, 2009b; Dhoke,
2009c).
New Approaches to Improve Clearcoat Scratch/Mar resistance

In recent researchers years have followed two different approaches to improve scratch
resistance: higher crosslink densities (UV curable and oligomeric clearcoats) and
nanoparticle reinforced clearcoats. In the first approach, increasing clearcoat hardness has
been aimed to be achieved (Ramezanzadeh, 2011c; Ramezanzadeh, 2011d, Tahmassebi,
2010; Dhoke, 2009a; Dhoke, 2009b; Dhoke, 2009c). It is well known that ceramic
materials have higher scratch resistance mainly due to their very high hardness. In fact,
materials which are hard enough prevent sharp objects penetrating into their surface. The
same idea has been used to achieve such properties in polymeric materials. Unlike
ceramic materials, polymeric substances are not hard enough to resists scratching objects
penetration. One way to increase polymeric clearcoats hardness is to use new generation
of clearcoats which have intrinsic higher cross-linking density, the most important of
which are oligomeric and UV curable clearcoats. Oligomeric clearcoats (which are
thermoset) have lower molecular weight precursors compared to the standard thermoset
clearcoats. These kinds of clearcoats are designed to achieve more regularly crosslinked

networks. It is thought that this network may improve scratch/mar performances of the
clearcoat noticeably. It has been shown that UV curable clearcoats have the potential to
increase scratch/mar resistance because they can achieve higher crosslink densities then
their typical thermoset counterparts. During UV-curing process, a liquid
photopolymerizable clearcoat mixture transforms to a 3D-solid network. This process can
be performed at room temperature and at very short times (resulting in saving energy and
time compared to other methods). It is also an environmental friendly process due to a
solvent-free polymerization method. High cross-linking density and 3D-structure of the
final cured clearcoat allow obtaining high scratch resistant films. However, there are
some weaknesses with the idea of increasing clearcoats scratch resistance by increasing
cross-linking density. Increase in crosslink density may lead to brittle coatings. This may
result in an increase in risk of cracking. Clearcoats having high crosslink density cannot
resist very high external forces. They behave like brittle materials when exposed to such
a condition. Therefore, a balance between brittleness and scratch resistance must be made
during formulation and development (Tahmassebi, 2010).
The addition of nanoparticles is another approach to increase the scratch/mar resistance

of the clearcoats. It is claimed that nano additives modify the viscoelastic properties of
clearcoats depending on their type, particles size and surface modification (Tahmassebi,
2010; Dhoke, 2009a; Dhoke, 2009b; Dhoke, 2009c; Zhang, 2007; Jalili, 2007). It seems
that such nano additives, compared with conventional additives, guarantee better
performances of a coating at extended periods of time. Nanoparticles owing to their small
particle size (<100 nm) are capable to produce strong physical/chemical interactions with
the coating matrix. The nature of interactions between clearcoat and particles mainly
depends on the surface modification and chemical composition of the clearcoat.
Nanoparticles depending on their primary size, surface modification and chemical
composition could vary clearcoats viscoelastic properties differently. It has been realized
that nanoparticles are able to influence clearcoats transition glass temperature (Tg), cross-
linking density, elastic modulus, hardness and toughness (Tahmassebi, 2010).The
simplest mechanism of improving scratch resistance of the clearcoat containing
nanoparticles has been their effect on increasing clearcoat hardness and crosslink density.
Inorganic nanoparticles like nano-clay, nano-SiO2, nano-Al2O3 and nano-ZnO have
hardnesses and elastic moduli greater than polymeric matrix of the clearcoat. Therefore,
they could increase coating resistance against scratching objects. However, depending on
the type of nanoparticles, different mechanisms may be influential on improving
clearcoat scratch resistance. It has been shown that some nanopraticles can reduce Tg and
hardness by reducing cross linking density and produce strong physical interactions with
the clearcoat matrix. In fact, nanoparticles may influence the clearcoat scratch resistance
by reducing the clearcoat brittleness and increasing its plastic flow. In this method,
although a high resistance cannot be achieved, the scratch morphology will not be in
form of fracture (Ramezanzadeh, 2011c; Tahmassebi, 2010; Dhoke, 2009a; Dhoke,
2009b; Dhoke, 2009c; Zhang, 2007; Jalili, 2007). Plastic type scratches are able to be
healed during service life at temperatures close to clearcoat Tg. Nanoparticles are also
found good toughening agents. They can increase clearcoat fracture energy.
It has been shown that the nanoparticles effects on the viscoelastic properties of a
clearcoat strongly depends on the nanoparticle dispersion. Incorporation of nanoparticles
into the clearcoat matrix needs specials techniques to prevent particles agglomeration. To
avoid such a problem, new methods like sol-gel process can be utilized to produce
nanoparticles in-situ (Ramezanzadeh, 2011e).

Sol-gel Process to Produce Scratch Resistance Clearcoats
It has been well understood that the homogeneous dispersion of the nanosized inorganic
particles within clearcoat matrix is one major task to avoid macroscopic phase separation.
The aggregation may occur during processing. This may influence physical/mechanical,
anticorrosion and chemical properties of the clearcoat. Moreover, using high loadings of
nanoparticles can undesirably increase coating viscosity, resulting in application
difficulties (Tahmassebi, 2010). In an in-situ process (sol-gel) of inorganic domain
formation in the clearcoat matrix, a good distribution of the inorganic phase with a strong
interaction with organic phase can be achieved. There are also other advantages of sol-gel
process including better mechanical properties of final formulation and also lower
increase in coating viscosity during application. The sol-gel process (bottom-up approach
of nanostructure formation) involves condensation and hydrolysis reactions of
appropriate organometallic precursors. These precursors, either as network former
(Ramezanzadeh, 2011e; Presting, 2003; Hernandez-Padron, 2003), such as tetraethyl
orthosilicate (TEOS) or as network modifier, such as methacryloxy
propyltrimethoxysilane (MEMO) and glycidoxy propyl trimethoxysilane (GPTS), can be
introduced to the main polymeric film former to obtain a so-called hybrid nanocomposite
film. Little studies have been carried out on the application of this approach in
automotive clearcoats resins. The overall film formation involves hydrolysis and
condensation of solgel precursor (Fig 8) as well as the ability of main polymeric matrix
to be cross-linked through appropriate functional groups in amino-cured acrylic polyol or
isocyanate-cured hydroxyl containing resins.
In addition, there is likely to observe an entanglement between organic functional groups

of solgel precursors such as acrylics in MEMO and epoxy in GPTS and the main film
former (Fig 9).
Fig. 8. Schematic stages of solgel precursor hydrolysis (1) and condensation (2a) and
(2b) (Ramezanzadeh, 2011f).
Based on such reactions, silicon containing hybrid nanocomposites are accessible,

enabling to obtain a mechanically reinforced system. The organic phase is responsible for
the general characteristics of the films such as adhesion, flexibility, and environmental
resistance of the coatings while the inorganic phase makes the film resist against
mechanical damages. There are many advantages for using solgel method to create
silica particles in a coating matrix (Ramezanzadeh, 2011e; Presting, 2003). The particle
size and distribution can be easily tuned by controlling the parameters affecting
hydrolysis and condensation reactions. An interpenetrating network can then be produced

in which organic and inorganic domains have been formed in nanometer size. However,
the presence of water to pursue the hydrolysis reaction may be a problem to carefully
adjust the miscibility of solgel precursors in solvent-based coating formulations.
Therefore, controlled hydrolysis and condensation reactions may prevent phase
separation, otherwise different types of morphologies may be obtained (Ramezanzadeh,
2011e).
Fig. 9: Schematic illustration of the reactions between acrylic polyol, melamine cross-
linker, and HTEOS (Ramezanzadeh, 2011f).
Self-healable Coatings
Self-healable coatings are those when are damaged and exposed to sunlight, are capable
to repair themselves. It has been shown that polyurethane coatings are inherently self-
healing due to their unique chemistry. In general view, polyurethane coatings show
typical properties listed below (Wall, 2009):
Hardness, toughness, and elasticity due to urethane and urea structure
High solids/low VOC

High reactivity and full crosslinking, even at low-temperature curing
Good chemical resistance and weather stability
Hydrogen bonds that allow for thermoplastic flow to relieve stress
and allow self-healing of defects
Basecoat compatibility
Pleasing optical properties a high gloss or wet look
It is well known that plastic type scratches can reflow when heated to a sufficient
temperature. On the other hand, fracture type scratches cannot be reversed by exposing
the clearcoat to elevated temperatures. In order to create a polyurethane system with high
self-recovery, the crosslink density should be increased while keeping the glass transition
temperature (Tg), low. In polyurethane system due to its high flexibility, the bonds are
stretched but not actually broken during marring (Wall, 2009; Ghosh, 2009). Therefore,
the bonds can return to their original shape when heated under sunlight. Schematic
illustration of scratch reflow is shown in Fig 10.
Fig 10: Schematic illustrations of scratch reflow when exposed to sunlight (Deretsky,
2011).
One way to produce polyurethane system with high self-healing capability is to use four-
molecule oxetane ring or a long rod of chitosan. Chitosan is closely related to chitin
which is a tough material obtained from shells of lobsters and crabs. The ring of oxetane
will be also ripped open when the polyurethane is damaged. The ultraviolet light can
easily enter the scratch through the broken surface of the polyurethane. The rod of
chitosan can be also broken inside the scratches. After that, the broken chitosan rods and
oxetane rings bond with each other across the damaged area. In this way, they can drag
the two sides of the scratch together to eliminate the scratch. It should be noted that the
reaction only takes place when the surface is damaged. In this way, it is expected to
produce polyurethane with complete scratch healing in an hour under sunlight (Wall,
2009; Ghosh, 2009; Deretsky, 2011, Syrett, 2011). The process can be easily seen in Figs
11 and 12.

Fig 11. Schematic illustrations of scratch self-healing process of the clearcoat containing
chitosan rods (Wall, 2009).
Fig 12: Schematic illustrations of scratch self-healing process of the clearcoat containing
chitosan rods (Deretsky, 2011).
Soft-Feel Coatings
In recent years, attempts have been carried out to use plastic materials in automotive
parts. In the car body, steel parts in the passenger compartment and in certain engine
components are wholly or partially replaced by plastic (Irle, 2008). The plastic parts also
need coating for making better appearance. The best choice for this purpose has been
soft-feel coatings. These coatings provide a luxurious look and leather-like texture to
hard plastic surfaces. They are mainly matt and can also improve the weather resistance
against sunlight and scratch. They are applied over the plastic substrates and are widely
used in automotive interiors. The coatings used in automotive industries for soft feel

applications have to meet the following requirements: (a) soft touch feeling and pleasant
handle, (b) resistance to chemicals, acids, bases, water, and other cleaners, (c) physical
durability-adhesion, impact, scratch resistance, abrasion and wear resistance, (d) low
gloss, (e) weather stability, (f) ambient or low temperature cure (Irle, 2008).
Two-component (2K) solvent borne polyurethane (PUR) has been used as a soft-touch
coating on plastics materials in the automotive industry. An appropriate soft-feel coating
requires the use of certain soft binders, e.g., aqueous hydroxy functional dispersions in
combination with water-emulsifiable polyisocyanates to formulate two-pack systems
(Weikard, 2007). It has been also understood that the combination of organic and
inorganic compounds can further enhance the soft-feel effect. Two different approaches
have been followed to prepare coatings with high soft-feel properties. First, surface
modifying additives can be used to modify the paint film texture and promote a "touch-
like" feeling. One of the surface modifying additives is wax. Waxes are surface
modifying agents which affect tactile property of the coating by changing surface
roughness and unevenness (Irle, 2008; Weikard, 2007; Martin, 2004; Stine, 2010;
Lockhart, 2004; Martin, 1998; Dannhorn, 1997). The conventional silicones, or
polydimethylsiloxanes (PDMS), are found suitable additives to improve soft-feel
properties of the coatings. These materials are highly surface-active materials due to the
arrangement of methyl groups on the flexible, inorganic backbone of alternating silicon
and oxygen atoms (Fig 13).
Fig 13: The general structure of conventional silicon additives for soft-feel application
(Irle, 2008).
These additives migrate to the surface of a cured film and improve lubricity (slip). This is
mainly because of the low intermolecular forces between the silicone polymer chains.
The second approach is to modify resin composition by using certain binders.
The required curing time to obtain fully curing is approximately cure cycles of 15 min
ambient and 30 min at 82 C.
One of the greatest problems of the soft-feel coating technology is its poor chemical
resistance. This is mainly because of the low cross-linking density of the soft-feel
coatings allows solvents and other aggressive reagents to permeate into the coating
matrix. One way to overcome this problem is increasing coating cross-linking density.
However, increasing the level of crosslinking reduces the softness of the film. New
findings show that the soft-feel coating resistance against chemicals can be improved by
increasing its cross-linking density in a UV curing method without reducing its soft-feel
properties. A UV-curable waterborne PUD can be obtained by incorporating polyester
acrylates into the non-functional waterborne PUD backbone for free radical

polymerization. A generalized structure of a UV-curable waterborne PUD is shown in
Fig 14.
There are many advantages in using UV-cure soft-feel coatings compared with traditional
one. The most important of which are better chemical and mechanical properties and
faster curing (curing time about 3 min).
Fig 14: Generalized structure of a UV-curable polyurethane dispersion.
New Trended For Soft-Feel Coating Technology

In recent years, VOC (volatile organic component) regulations have resulted in a shift in
soft-feel coating chemistry toward waterborne coatings. In this regard, waterborne
polyurethane polyurethaneAny of a class of very versatile polymers that are made into
flexible and rigid foams, fibres, elastomers (elastic polymers), surface coatings, and
adhesives formulations (with low volatile organic compounds)volatile organic
compound Environment Any toxic cabon-based (organic) substance that easily become
vapors or gaseseg, solventspaint thinners, lacquer thinner, degreasers, dry cleaning
fluids have become the choice for the soft feel coatings. Therefore, traditional solvent
borne additives (insoluble in water) cannot be used as softening agents. Therefore, they
cannot be used in waterborne coatings. Attempts have been made to modify silicon
additives in order to obtain compatibility with waterborne coatings. This includes
physical modifications, pre-dispersing the PDMS as an emulsion, a self-dispersing PDMS
compound, chemical modification and grafting hydrophilic groups onto the PDMS chain.
Four common types of waterborne polyurethane coating formulations (with different
combinations of polymer resins) Versatlie liquid plastic coating. It self-levels and hardens
to produce a thick, clear, durable and glossy finish. have been utilized for the soft feel
applications. These include waterborne polyester polyols, polyurethane dispersions
(PUD), combination of polyester polyol with PUD, and combination of acrylic polyol
with PUD. The first sot-feel waterborne coating can be obtained as combination of a
dispersible polyester polyol with a hydrophilic aliphatic isocyanate. The second approach
can be gained by formulating polyurethane dispersions with hydrophilic polyisocyanate.
The third type waterborne coating with soft-feel application can be obtained by adding
hydrophilic polyisocyanate (as cross-linker) to the mixture of polyester polyol and PUD.
The forth type can be formulated by adding hydrophilic polyisocyanate to the mixture of

acrylic polyol with polyurethane dispersions (Irle, 2008; Weikard, 2007; Martin, 2004;
Stine, 2010; Lockhart, 2004; Martin, 1998; Dannhorn, 1997).
Self-cleaning Coatings
Self-cleaning surface is a specific type of surface which keeps itself free of dirt and
grime. There are two main approaches to create a self-cleaning surface. Making the
surface superhydrophobic or turning it to a superhydrophilic surface. In both approaches,
self-cleaning is performed through the action of water; the former by rolling droplets and
the latter by sheeting water that carries away dirt.
Superhydrophobic Surfaces
The idea of creating this kind of self-cleaning coatings refers to the lotus leaf. SEM
images revealed that the key features of the lotus leaf were a microscopically rough
surface composing of an array of randomly distributed 5-10 mm micropapillars(Fig. 15
from ref. (Wang, 2011) ). These micropapillars were covered with a waxy hierarchical
nanostructure. The water contact angle on a lotus leaf was higher than 160 with a rolling
angle of about 2. Silver ragwort leaves, water striders leg, rice leaves and waterfowl
feathers have the similar structures and can be called superhydrophobic surfaces (Wang,
2011). The contaminants on such surfaces are swept by water droplets or adhere to the
water droplet and are removed from the surface when the water droplets roll off (Fig. 16
, reproduced from ref. (Crick, 2010)). The combination of low surface energy materials
together with micro- and/or nano-structured features, which can potentially reduce the
contact area between the surface and water droplets, form superhydrophobic surfaces.
Fig. 15. a) A few water droplets on a superhydrophobic surface and (b) low and (c) high
magnification images of a typical lotus leaf, reproduced from ref. (Wang, 2011)).
a b c
The least requirements for a hydrophobic self-cleaning surface are very high static water
contact angle , often quoted >160, and a very low hysteresis angle, the minimum
inclination angle necessary for a droplet to roll off on the surface. In order to give such
characteristics to a surface, surface free energy of the surface must be reduced and
roughness has to be increased. To this end, application of low-surface-energy materials
such as fluorinated- or siliconized- based polymers can be used (Ma, 2006; Shirtcliffe,
2011; Henderson, 2009; Yan, 2011). Up to now, various techniques such as Layer-by-
Layer Deposition (Zhai, 2004), Electrodeposition/electropolymerization (Darmanin,
2010), Plasma and Laser Treatment(Pennisi, 2010), Electrospinning(Jin, 2011), Casting
and Molding (Dodiuk,2007), have been implemented to elaborate such micro,
nanostructures.

Fig. 16. The self-cleaning action of a superhydrophobic surface when a water rolls off on
it, reproduced from ref. (Crick, 2010).
In the first products, although these surfaces were effectively self-cleaning, they had a
number of short comings which had prevented their widespread applications. Preparation
of such surfaces was kind of costly and time consuming and the resultant coatings were
usually hazy which had restricted their applications on lenses and windows. In recent
studies and products, these drawbacks have been eliminated. For example, a new product,
called Nanoshell, which contains silane based nano particles, is a clear, UV resistant,
non-stick, easy to clean coating. This coating is recommended for glass and ceramic
surfaces, especially for windshield glasses of automobiles. This coating is resistant
against bird droppings, tree saps and pollutions such as acid rains too.
Gimat has engineered nanocoatings on glass that not only does not reduce the clarity of
the substrate, but also the transmittance of the coated glass is actually improved with
respect to the bare substrate due to the anti-reflective nature of the coating. This novel
coating would be considered a high performance superhydrophobic self-cleaning surface.
Superhydrophilic Surfaces
Hydrophilic surfaces are based on some photoactive materials such as titania (Banerjee,
2011; Fujishima, 2008). They can chemically break down adsorbed dirt in sunlight. This
class of self-cleaning surfaces can remove any organic pollutions and dirt in two stages.
In the first stage, when the coating is exposed to an especial part of electromagnetic
spectrum (usually UV part), any organic dirt in contact to the coating is chemically
broken down. This process is known as photocatalysis(Fig. 17). As a result of
photocatalytic reactions, the surface is converted from a hydrophilic surface to a
superhyrophilic one. In the second stage, unattached dirt can be washed away by a flow
of water (such as rain), because water spreads evenly on superhydrophilic surfaces with
contact angles reaching zero.
Among different photocatalytic materials, titania is the most popular option because it is
highly efficient at photocatalyzing dirt in sunlight. In recent years, different glass
manufacturers such as Pilkington, PPG industries, Cardinal Glass Industries, acclaimed
that they could commercialize the production of self-cleaning glasses. All of these
windows had been coated with a thin transparent layer of titanium dioxide. For example,
PPG's SunClean Windows are "self-cleaning" and contain an external thin film coating
containing nanoparticles.

Fig. 17. The mechanism (left) and the self-cleaning action of a superhydrophilic surface
when a water evenly spread on it reproduced from ref. (Fujishima, 2008).
Despite the commercialization of hydrophilic self-cleaning coatings on glass in a number

of products, creating such coatings on organic substrates has not achieved considerable
successes. One of the most important applications for such coatings is the car body
coating systems. In this case, photocatalytic degradation of organic substrate i.e. the
automotive topcoat in contact with the photoactive self-cleaning coating is a great
challenge. Although in recent studies (Carneiro, 2007; Lee, 2009, Fateh, 2011), some
experimental techniques for obviating this problem has been given, the
commercialization of these novel coatings is far from maturity.
Anti-reflecting coatings
Reflection, as a fundamental optical phenomenon, is inevitable when light reaches a
boundary between two media having different refractive indices. In many applications a
strong reflection from a smooth surface is unpleasant, e.g. for displays or spectacles. The
compound eye of some insects such as bees has a unique structure in the cornea that
reduces reflectivity. This structure (Fig. 18) has been a perfect pattern for development of
anti-reflective coatings ( Min, 2008, Li, 2010).
Reflection can be lowered by reducing the refractive index differences between the two
media (Fresnel equation). The reflectivity of a surface can be reduced by two different
concepts: the first describes a micro- and nano-structure, which reduces uniform
reflection through light scattering. This concept is an imitation of nano-moth eye
structure. The second concept is based on destructive interference of the reflected light at
the upper and the lower interfaces of the thin film/substrate system (interference). In
order to create interference type anti reflecting coatings, materials of different refractive
indices must be used alternatively (Qingna, 2007). To this end, materials possessing high
(such as titania) and low (such as silica) refractive indices can be utilized. These two
materials provide a strong distinctness in refractive index leading to a significantly higher
transmission and reduced reflection. In the interference system the coating system can be
single layer or multi-layer that can perform on a very narrow or a broad band of incident
light, respectively. The thickness of the layers is in the range of several hundred
nanometers. Although the second concept is more expensive, it offers a better control.

Fig. 18. Macroscopic image (left) and Fine detail (right) on Moth's Eye as a model for
anti-reflective coating Surface (.Laser Focus World, Aug. 1999 Pg 109).
Anti-reflecting coatings have been used for inorganic substrates for several decades
(Prado, 2010). In recent years, many researchers have been studying to use these coatings
on the organic substrates such as plastics (Kleinhempel, 2011). It requires replacing
conventional depositing methods like vacuum based ones with novel processes such as
sol-gel based methods (Zalga, 2008; Muromachi, 2006). The main advantage of sol-gel
methods is the lower processing temperature that avoids deformation of organic
substrates. The coatings resulted from this process have good mechanical, optical,
chemical and environmental properties meeting the requirements of the automotive
industry. For example, these coatings have been successfully commercialized and applied
as the cover plate for the dashboard display in Toyota Prius. In other attempts solgel
process has been widely employed to prepare antireflective coatings on different glasses
such as automotive windows (Muromachi, 2006).
In a new research done by several materialists from Institut fr Neue Materialien at
Saarbrcken, Germany, they could produce a novel anti-reflecting coating using SiO2
and TiO2 nanoparticles of 10 nm and 4 nm respectively to minimize reflections from the
dashboard into the drivers view. DSM has already launched Claryl picture glass, which
reduces the amount of reflection causing framed pictures to be seen clearer. Claryl is
currently available in 15 European countries and has also recently been launched in the
USA.
Anti-fog Coatings
Bringing a cold surface into a warmer surrounding environment will lead to fogging. This
effect is inevitable unless the surface is heated. This is a consequence of formation of tiny
droplets on the mirrors surface scattering the light. A superhydrophilic coating can
prevent droplet formation to a certain extent. The droplets simply merge into a thin water
layer on the mirror without changing the transmission too much. Like superhyrophilic
self cleaning coatings, photocatalytic TiO2 coatings are super-hydrophilic when exposed
to sufficient UV light (Fig. 19 , (reproduced from ref. (Fujishima, 2008))). The material
prevents fogging by not allowing water to form beads on surfaces. The first anti fogging
coating, commercialized in Japan, was a novel coating for automobile side-view mirrors.
The TOTO Ltd., is currently supplying TiO2-treated glass for major automobile
manufacturers.
Extending the response wavelength of titanium into the visible light region is the main
field of researches for the design and actual applications of these novel type of coatings.

Fig. 19. Photographs of fogged surfaces of typical glass (left) and TiO2-coated glass
(right) after exposure to UV (reproduced from ref. (Fujishima, 2008))
Miscellaneous
Along with technologies given in preceding parts, there are other potential technologies
in which promising perspectives are forecasted for them. Some of these are briefly
presented.
Energy-saving and comfort: by controlling the solar energy passed via windows and
entered the car, a dramatic increase in energy efficiency is achieved. Low-E and
dichromatic multiple coatings are of the latest energy-saving systems which avoid
transmission of unwanted UV and IR radiation thorough reflection and absorption of
these wavelengths, respectively. Sputtered low-E glasses are currently in use for
automotive applications such as windshields.
Dichromatic systems, which have been recently developed, are smart and switchable.
They have this ability to control certain features of windows and mirrors so that they
darken selectively on command. They can be used to provide glare control, daylight
management, and reduced energy consumption. Electrochromic coatings are the most
typical materials among various dichromatic technologies. These multi-layer systems
include different conductive layers. One of these layers is an active electrochrome which
can switch between its oxidized and reduced states as a result of redox
reactions(MONK,2007). These alterations cause a change in color or grayness of the
electrochromic system. Electrochromic automotive windows and mirrors are the largest
application of these materials today. In these applications, the mirror darkens
automatically at night with the approach of bright headlights.
These electrochromic mirrors and windows are typically found on high-end cars.
Unfortunately, the complex structure of electrochromic systems leads to high
manufacturing costs. Another limitation for such systems is their response time which is
too slow to be widely used in some applications. By mass production of electrochromic
devices which reduces the cost together with some recent findings in which a significant
improvement in response time has obtained, a bright future for electrochromic devices is
viewed.
Smart coatings: The NASA Corrosion Technology Laboratory has developed a novel
approach to fight rust and corrosion on metals by the aid of a smart coating. These
innovative coatings which are called self-healing coatings, are of interest to be used in
automobile painting. These coatings which have been the target of numerous studies,
millions of micro- or nano-capsules are embedded in paint to inhibit rust (Fig. 20). The

chemicals, which are usually corrosion inhibitors, stay inside the capsules until the
corrosion reactions begin. The pH variations resulted from corrosion reactions cause the
capsule shell to be torn and corrosion inhibitors are released, preventing corrosion
(Baghdachi, 2009; Lvov, 2005). In another type of smart coatings with the same
mechanism, capsules are filled with materials that change the color of the affected area.
When the corrosion begins, as a consequence of such color variation, corrosion is
highlighted. The researches in the field of smart coatings are ongoing.
Fig. 20. Schematic illustration of smart coatings action when corrosion begins.
Iridescent colors / photonic crystals: The wings of some butterflies, which have a
regular pattern of holes, or nacre, or mother of pearl, which is a layered structure
composed of hexagonal platelets and precious opals have been the natural models for
creating interference colors. The same structure can be artificially manipulated. In a new
innovative approach, silica spheres are arranged in a regular pattern which provides the
base for complex light interference.
Weathering resistance nanocoatings: In recent researches, various nanoparticles have

been incorporated into conventional polymeric coatings to enhance their resistance
against sunlight. For this purpose, some nano particles such as zinc oxide, iron oxide,
cerium oxide, titanium oxide and silica can be utilized (Peng, 2008; Oliveira, 2008).
These nanoparticles have this ability to absorb in the harmful part of sunlight (ultraviolet
part), preventing the polymeric coatings from weathering degradation.
REFERENCES
ACT Test Panels, Inc.; Michigan, USA.
Amerio E., Fabbri P., Malucelli G., Messori M., Sangermano M., Taurino R., Scratch
resistance of nano-silica reinforced acrylic coatings, Prog. Org. Coat. 62 (2008) 129.
Baghdachi J, Smart Coatings, Smart Coatings II, , ACS Symposium Series, Volume 1002
Chapter 1, 2009, pp 3-24.
Bajata J.B., Mikovi-Stankovi V.B., Popi J.P., Drai D.M., Adhesion characteristics
and corrosion stability of epoxy coatings electrodeposited on phosphated hot-dip
galvanized steel, Prog. Org. Coat. 63 (2) (2008)
Bajat J.B., Miskovi'c-Stankovi'c V.B., Bibi'c N., Drazi'c D.M., The influence of zinc

surface pretreatment on the adhesion of epoxy coating electrodeposited on hot-dip
galvanized steel, Prog. Org. Coat. 58 (2007)323.
Banerjee AN, The design, fabrication, and photocatalytic, utility of nanostructured

semiconductors: focus on TiO2-based nanostructures, Nanotech, Sci Appli 4, 2011, 35.
Barberich B (2004) Application of Color Powder Paint in the Automotive Indus-try;

DeWitt CM.
Barletta M., Bellisario D., Rubino G., Ucciardello N., Scratch and wear resistance of
transparent topcoats on carbon laminates, Prog. Org. Coat. 67 (2010) 209.
Bautista Y., Gmez M.P., Ribes C., Sanz V., Correlation between the wear resistance,
and the scratch resistance, for nanocomposite coatings Prog. Org. Coat., 70 (4) (2011)
178-185.
Biller (2006) OEM Automotive Powder Coatings.Section 1: Scratch/wear resistance

clearcoats. (1995) A Holistic Approach to Automotive Powder Coating; and Kevin
Beck, E. Into the third dimension: Three ways to apply UV coating technology to 3-D
automotive objects, Euro. Coat. Jour. 4, 2006, 32-39.
Bell G. , Powder coatings for highly specified automotive applications, Surf Coat Inte
Part B: Coat Transa, 81, 1998, 485.
Berger D. M., Wint R. F., ASTM Special Technical Publication 841; American Society
for Testing and Materials; 1984
Biller, K. (2006). OEM Automotive Powder Coating, PCI Magazine Available online
at:
http://www.pcimag.com/Articles/Feature_Article/cbf6aeb98e85c010VgnVCM100000f93
2a8c0 (Ac-cessed 10/27/2009).
Carneiro J.O., Teixeira V., Portinha A., Magalhes A., P. Coutinho, C.J. Tavares, R.
Newton, Newton Iron-doped photocatalytic TiO2 sputtered coatings on plastics for self-
cleaning applications, Mate Sci Eng, 138, 2007,144.
Cho K.W., Rao V.S., Kwon H., Microstructure and electrochemical characterization of
trivalent chromium based conversion coating on zinc, Electrochim. Acta 52 (2007) 4449.
Corrigan V., Zawacky S., High Edge Coverage Electrocoat Development for Enhanced
Corrosion Protection US Patent 5,096,556, 1992.
Courter J.L., Mar resistance of automotive clearcoat: I. Relationship to coating

mechanical properties, J. Coat. Technol. 69 (866) (1997) 57.
Crick CR. and Parkin IP., Preparation and Characterisation of Super-Hydrophobic

Surfaces, Chem. Eur. J. 2010, 16, 3568.
Dannhorn W., Patent US 5,684,081 (1997).
Darmanin, T.; de Givenchy, E. T.; Amigoni, S.; Guittard, F. Langmuir 26, 2010, 17596.

Deretsky Z., National Science Foundation, after Burnworth et al., Nature, April 21, 2011.
Di Maggio R., Fedrizzi L., Rossi S., Jour.Adh. Sci.Techn. 15(7) (2001) 793.
Dhoke S.K., Khanna A.S., Electrochemical behavior of nano-iron oxide modified alkyd
based waterborne coatings, Mater. Chem. Phys. 117 (2009) 550-556.
Dhoke S.K., Khanna A.S., Sinha T. Jai M, Effect of nano-ZnO particles on the corrosion
behavior of alkyd-based waterborne coatings, Prog. Org. Coat. 64 (2009) 371-382.
Dhoke S.K., Khanna A.S., Effect of nano-Fe2O3 particles on the corrosion behavior of
alkyd based waterborne coatings, Corros. Sci. 51 (2009) 6-20.
Dodiuk H., Rios P. F., Dotan A., Kenig S., Hydrophobic and self-cleaning coatings,
Polym. Adv. Technol. 2007; 18: 746.
Fateh R, Ismail AA., Dillert R, Detlef W. J. , Highly Active Crystalline Mesoporous

TiO2 Films Coated onto Polycarbonate Substrates for Self-Cleaning Applications, Phys.
Chem. C, 115, 2011,10405.
Fedrizzi L., The use of electrochemical techniques to study the corrosion behaviour of
organic coatings on steel pretreated with solgel zirconia films Electrochimica Acta,
46(24-25) (2001) 3715.
Fujishima A, Zhang X, Tryk DA., TiO2 photocatalysis and related surface

phenomena,Surf Sci Rep 63 (2008) 515.
Ghosh B., Urban M. W. Science. 323 (13) (2009) 1458.
Guo Z, Liu W, Su BL, Superhydrophobic surfaces: From natural to biomimetic to

functional, Jour Collo Inte Sci 353 (2011) 335.
Gusmano G., Zirconia primers for corrosion resistant coatings, Surf. Coat. Technol.,
201(12) (2007) 5822.
Hara Y., Mori T., Fujitani T., Relationship between viscoelasticity and scratch
morphology of coating films. Prog. Org. Coat. 40 (2000) 39.
Harun M.K., Marsh J., Lyon S.B., The effect of surface modification on the cathodic
disbondment rate of epoxy and alkyd coatings Prog. Org. Coat. 54 (4) (2005) 317.
Henderson, Karen. Self-healing polyurethane clearcoats: new developments for exterior

automotive coatings. Paint & Coat Indu, 25, 2009 40.
Hernandez-Padron G., Rojas F., Garcia-Garduno M., Canseco M.A., Castano V.M.,
Development of hybrid materials consisting of SiO2 microparticles embedded in
phenolic-formaldehydic resin polymer matrices, Mater Sci Eng A, 355 (2003) 338.
Hirai S., Surface Pretreatment Based On Dilute Hexafluorozirconic Acid,

Jour.Amer.Ceram. Soci., 81(12) (1998) 3087.

Jalili MM, Moradian S, Dastmalchian H, Karbasi A. Investigating the variations in
properties of 2-pack polyurethane clear coat through separate incorporation of
hydrophilic and hydrophobic nano-silica, Prog. Org. Coat, 59 (2007) 81-87.
Jardret V., Lucas B.N., Oliver W., Scratch Durability of Automotive Clear Coatings: A
Quantitative, Reliable and Robust Methodology, J. Coat. Technol. 72 (907) (2000) 79.
Jin, G.; Kim, G. Langmuir 27, 2011, 828.
Lee CS, Kim J, Son J.Y., Choi W, Kim H, Photocatalytic functional coatings of TiO2
thin films on polymer substrate by plasma enhanced atomic layer deposition, Appli
Cataly B: Environmental, 91, 2009, 628.
Li Q., Xu S., Hu J., Zhang S., Zhong X., Yang X., The effects to the structure and
electrochemical behavior of zinc phosphate conversion coatings with ethanolamine on
magnesium alloy AZ91D, Electrochim. Acta. 55 (2010) 887.
Li Y, Zhang J, Yang B, Antireflective surfaces based on biomimetic nanopillared arrays,

Nano Today (2010) 5, 117.
Li H.B., Corrosion protection of mild steel by zirconia sol-gel coatings, Journal of

Materials Science Letters, 20(12) (2001) 1081.
Lister J., ICI Paints; 1992
Lockhart A, Gindin L, Schmitt P, Lunney P, Technical paper, Radtech Report,

November/December 2004.
Maag, K., Lenhard, W., Loeffles, Helmut, New UV curing systems for automotive
applications, Prog. Orga. Coat., 40, (2000) 93.
Marinovic S. , Popovic I. , Dunjic B. , Tasic S. , Bozic B. , Jovanovic D. , he influence of

different components on interpenetrating polymer network's (IPN's) characteristics as
automotive top coats, Prog. Orga. Coat. 68 (2010) 293.
Marsh J., Scantelbury J.D., Lyon S.B., The effect of surface/primer treatments on the
performance of alkyd coated steel, Corros. Sci. 43 (2001) 829.
Martin K., Adamsons K., Henderson B., Chemical attack in automotive interiors
Barrier Properties of 2K Coatings, 2004.
Ma M, Hill RM., Superhydrophobic surfaces, Current Opinion in Colloid & Interface

Science 11 (2006) 193.
Martin K., Shaffer M., Advanced Coatings Technology Conference Proceedings, Detroit,
MI, Sept. 1998,
Meine, D., More light - Less energy, Euro. Coat. Jour, 12, (2007) 54.
McMillan R. E., Electrocoat 2002 Conference, Orlando, FL; The Electrocoat Association,
Cincinnati, OH, 2002; pg. 151.

Min WL, Jiang B, Jiang p, Bioinspired Self-Cleaning Antireflection Coatings, Advan
Mate 20, 2008,1.
MONK P. M. S., MORTIMER R. J., ROSSEINSKY D. R., Cambridge University Press,

2007, 395-416.
Muromachi T, Tsujino T, Kamitani K, Maeda K (2006) J SolGel Sci Technol 40:267

272.
Narayanan S., Surface pretreatment by phosphate conversion coatings- A review, Adva.

Mater. Sci., 9(2) (2005) 130.
Oliveira C.F.P., Carastan D.J., Demarquette N.R., Fechine G.J.M., Photooxidative

Behavior of PolystyreneMontmorillonite Nanocomposites, Polym Eng Sci , 2008.
Orzech, S, Pagac, E, Next-generation powder clearcoats, Finishing Today, 83, 2007, 34.
Peng G., Yang Li Q., Y., H., Wang LW., Effects of nano ZnO on strength and stability of
unsaturated polyester composites, Polym Adva Techno, 19 (2008)1629.
Pennisi, C. P.; Zachar, V.; Gurevich, L.; Patriciu, A.; Struijk, J. J.; Conf. Proc. IEEE Eng.
Med. Biol. Soc. 2010, 2010, 3804.
PPG Industries, Inc. Industrial Coatings.
Prado R, Beobide G, Marcaide A, Goikoetxea J, Aranzabe A, Development of

multifunctional solgel coatings: Anti-reflection coatings with enhanced self-cleaning
capacity, Sol Ener Mate Sol Cell, 94, 2010,1081.
Presting H, Konig U, Future nanotechnology developments for automotive applications,

Mater Sci Eng C, 23(68) (2003) 737.
Kleinhempel R, Wahl A, Thielsch R, Large area AR coating on plastic substrate using

roll to roll methods, Surf Coat Tech, 205, 2011, 502.
Qingna Z, Yuhong D, Peng W, Xiujian Z, CeO2-TiO2/SiO2 Anti-Reflecting and UV-

Shielding Double-Functional Films Coated on Glass Substrates Using Sol-Gel Method, J
Rare Earths, 25, 2007, 64.
Ramezanzadeh B., Attar M.M., Farzam M., Corrosion performance of a hot-dip

galvanized steel treated by different kinds of conversion coatings, Surface & Coatings
Technology 205 (2010a) 874.
Ramezanzadeh B., Attar M.M., An evaluation of the corrosion resistance and adhesion
properties of an epoxy-nanocomposite on a hot-dip galvanized steel (HDG) treated by
different kinds of conversion Surface & Coatings Technology 205 (2011b) 4649.
Ramezanzadeh B., Moradian S., Tahmasebi N., Khosravi A., Studying the role of
polysiloxane additives and nano-SiO2 on the mechanical properties of a typical
acrylic/melamine clearcoat, Prog. Org. Coat, 72 (4) (2011) 621-631.

B. Ramezanzadeh, S. Moradian, A. Khosravi, N. Tahmasebi, A new approach to
investigate scratch morphology and appearance of an automotive coating containing
nano-SiO2 and polysiloxane additives Prog. Org. Coat, 72 (3) (2011) 541-552.
Ramezanzadeh B., Moradian S., Yari H., Kashani A., Niknahad M., The effect of
basecoat pigmentation on the scratch resistance and weathering performance of an
acrylicmelamine basecoat/clearcoat automotive finish, Tribology International,
doi:10.1016/j.triboint.2011.10.016.
Ramezanzadeh B., Moradian S., Khosravi A., Tahmassebi N., Effect of polysiloxane
additives on the scratch resistance of an acrylic melamine automotive clearcoat, J. Coat.
Technol. Res. DOI 10.1007/s11998-010-9239-4.
Ramezanzadeh B., Mohseni M., Karbasi A., Preparation of solgel-based nanostructured

hybrid coatings; part 1: morphological and mechanical studies, J Mater Sci, DOI
10.1007/s10853-011-5819-7.
Roberts and Caserio; Basic Principles of Organic Chemistry; Benjamin; 1964

Saravanan G., Mohan S., Corrosion behavior of Cr electrodeposited from Cr(VI) and
Cr(III)-baths using direct (DCD) and pulse electrodeposition (PED) techniques, Corros.
Sci. 51 (2009) 197.
Schmael W. R., Purcell J., (eds); New Polymer Technology for Auto Body Exteriors,
AIChE Symposium Series, No. 260, Volume 84; American Institute of Chemical
Engineers; 1988.
Schmidt, H., Fink, D. Latest developments with acrylic powder clearcoat for automotive
OEM, JOCCA - Surface Coatings International, 79, 1996.
Schulz U., Wachtendorf V., Klimmasch T., Alers P. The influence of weathering on
scratches and on scratch and mar resistance of automotive coatings, Prog. Org. Coat, 42
(2001) 38.
Schwarzentruber, P., Scratch resistance and weatherfastness of UV-curable clearcoats,

Macromolecular Symposia, 187, 2002, 531.
Seubert, C.M., Nichols, M.E., Alternative curing methods of UV curable automotive

clearcoats, Prog. Orga. Coat. 49 (2004) 218.
Simi S. , Dunji B. , Tasi S. , Boi B. , Jovanovi D. , Popovi I. , Synthesis and

characterization of interpenetrating polymer networks with hyperbranched polymers
through thermal-UV dual curing , Prog. Orga. Coat., 63, 2008, 43.
Shen W., Jiang B, Scholten A., Schwenke R., Mi L., Seal C., Wang P., A Quantitative
Index for Mar and Scratch Resistance of Materials for Automotive Glazing Applications
and Quantitative Evaluation of Damages by Different Scratching Modes, Tribogy Letters,
17(3) (2004) 637-644.
Shen W, Bin J., Steven M.G., Harold M., Study of tribological properties of
coating/substrate system in micrometer and nanometer scales with a scanning probe
microscope, Tribol. Int. 34 (2001) 135.

Shirtcliffe NJ., McHale G, Newton MI., The Superhydrophobicity of Polymer Surfaces:
Recent Developments, JOUR POLY SCI PART B: POLY PHYS 49, 2011, 1203.
Shutic, J.R., process and application trends in automotive powder coating systems
relevant to "Class A" requirements, Technical Paper - Society of Manufacturing
Engineers. 98, 1998, Pages X1-27.
Snuparek, J., Kadrnka, B., Ritz, P., Quadrat, O., UV stabilisation of powder clear coats,
Macrom Sympo , 187, 2002, 343.
Srensen P.A., Kiil S., Dam-Johansen K., Weinell C.E., Influence of substrate
topography on cathodic delamination of anticorrosive coatings, Prog. Org. Coat. 64
(2009) 142.
Stine C. , Orvis E. , Yousif N. Z. , Paquet D. A. , Sample K.M. , Yahkind A.L., Patent

No 20100092766, (2010).
Stropp, J.P., Wolff, U., Kernaghan, S., Lffler, H., Osterhold, M., Thomas, H., UV curing
systems for automotive refinish applications, Prog. Orga. Coat., 55,(2006) 201.
Syrett J.A., G.Mantovani, Barton W.R. S., Price D., Haddleton D. M., Polymer
Chemistry. DOI: 10.1039/b9py00316a.
Tahmassebi N., Moradian S., Ramezanzadeh B., Khosravi A., Behdad S., Effect of
addition of hydrophobic nano silica on viscoelastic properties and scratch resistance of an
acrylic/melamine automotive clearcoat, Tribo Intern, 43(2010) 685.
Thorstenson T.A., Huang J.B., Urban M.W., Mobility and distribution of silicone
additives in coatings; a spectroscopic study, Prog. Org. Coat. 24 (1994) 341.
Tyler, S, UV technology can make a difference, Polym. Paint Colo Jour
200, 2010, 54.
Valet, A., Rogez, D. , Light stabilisation of UV-curable powder coatings, JOCCA - Surf
Coat Intern, 82, 1999, 293.
Wall M. , Discover Magazine, http://blogs.discovermagazine.com
Wang X, Ding B, Yu J, Wang M, Engineering biomimetic superhydrophobic surfaces of

electrospun nanomaterials, Nano Today (2011) 6, 510.
Wazer J.R.V, Vol 2, 1967, Interscience Publishers Inc.: New York.
Weikard J., The New Frontier: Waterborne UV Coatings for Plastics, 2007.
Yan YY, Gao N., Barthlott W., Mimicking natural superhydrophobic surfaces and
grasping the wetting process: A review on recent progress in preparing superhydrophobic
surfaces, Advan in Collo Interf Sci 169 (2011) 80.
Yari H., Moradian S., Ramazanzade B., Kashani A., Tahmasebi N., The effect of
basecoat pigmentation on mechanical properties of an automotive basecoat/clearcoat
system during weathering, Polym Degrad Stab 94( 2009) 1281.

Yuri ML, Shchukin DG., Mohwald H, Price RR., Halloysite Clay Nanotubes for
Controlled Release of Protective Agents, ACS Nano, , 2 (5), 2008, 814820.
Zalga A , Kazadojev I , A Beganskiene , Kareiva A , Sol-Gel Synthesis of Modified

Silica Anti-Reflecting Coatings, NATO Science for Peace and Security Series C:
Environmental Security, NATO Science for Peace and Security Series C: Environmental
Security, 2008, Sol-Gel Methods for Materials Processing, Pages 489-495
Zhang X., Wang F., Yuanlong D., Effect of nano-sized titanium powder addition on
corrosion performance of epoxy coatings, Surf. Coat. Technol. 201 (2007) 7241-7245.
Zhai, L.; Cebeci, F. C.; Cohen, R. E.; Rubner, M. F. Nano Lett. 2004, 4, 1349.

Automotive Lower Suspension Arm: Experimental and
Simulation Durability Assessment
N.A. Kadhim, N. Nikabdullah, S. Abdullah and A.K. Ariffin
ABSTRACT
In engineering, there is a type of failure which is called fatigue and caused by the
repeated loading of components. This type of failure has been investigated under service
signals for one of the critical automotive components represented by the lower
suspension arm. The characteristics of original and edited signals were statistically
studied in time domain then both of the signals used for component prediction purposes.
The fatigue tests were conducting for the SAE 1045 steel based strain-controlled. A
fatigue test under constant amplitude loading has been carried out to specify the material
mechanical properties of the used material then the results were provided for the finite
element code. The finite element fatigue life were evaluated and compared with the
experimental results. The accuracy and efficiency of the numerical models are
demonstrated. Finally, it is suggested that the approach presented in this chapter can be
used for initial fatigue analysis and more suitable fatigue prediction models under service
loadings need to be developed.
INTRODUCTION
The subject of this study is the lower suspension arm that serves as the locators for the
suspension system (Devlukia and Bargmann, 1997; Haiba et al., 2003; Kim et al., 2003;
Mrzyglod and Zielinski, 2007). The lower suspension arms are attached to the vehicle
frame with bushings and allow the wheels to move up and down in response to the road
surface (Erjavec, 2006). Structural components such as a lower suspension arm might be
strong enough to withstand a single applied load. But what happens when the part
operates over and over, day after day? To predict component failure in such cases
requires what is called fatigue or durability analysis.
Fatigue analysis is one of the key stages in the automotive design for which the durability
testing is extensively performed in most of related industries. Much of this testing
involves simulating field or service loading using a complex variable amplitude histories
which is considered as one of the vital input. These histories may be from the respective
industrial standards, such as the FALSTAF or TWIST histories, which are normally used

in the aircraft industry (Stephen, 1997). In addition, some histories were generated by
instrumenting components and vehicles which are operated under varying service
conditions.
With the introduction of finite element analysis (FEA) techniques, however, it has come
to the possibility of performing fatigue calculations prior to the development of a
prototype. In classic structural analyses, failure predictions are solely based on the
material strength or the yield strength. The durability analysis goes beyond this, and the
estimation of fatigue life is based on a repeated simple or a complex loading.
The finite element (FE) based durability analysis can be considered as a complete
engineering analysis for the components. The fatigue life can be estimated for every
element in the FE model, and the contour plot of life or damage. The FEA results define
the stress-state for a component given the specific loading condition. The most common
FEA method used in conjunction with the fatigue analysis is to apply each load
independently as a unit load case. The inputs for the FEA are the component geometry
with FE model, the boundary conditions, and the loading information.
Though many low-cycle fatigue data have been published with or without FEA, some of
them for automotive components especially the lower suspension arms (Haiba et al.,
2002; Haiba et al., 2003; Song et al., 2009) have been conducted with different attitudes.
A work from Haiba et al. (2002) has been estimated the fatigue lives of lower arm in both
time and frequency domain methods under FEA. Another work by Haiba et al. (2003)
introduced a new structural optimisation algorithm based on fatigue life. The work
investigated the effects of different assessment strategies on the predicted fatigue life of a
lower suspension arm. Recently, Song et al. (2009) performed a simulation based FE
model analysis for an automotive control arm. The fatigue analysis is carried out to verify
the final design durability. The suggested optimum design considering strength has been
investigated in terms of durability performance using strain life method.
This work is aimed for the automotive industry, more specifically the fatigue estimation
of the part which is related to a suspension system, where safety is a great concern. The
fatigue lives were estimated using the available models in the FE package under original
and edited service signals. The results show a big difference between experimental and
FEA fatigue estimation life. The results reflect the need for new models associated with
FEA package which can be more suitable to predict fatigue life under variable amplitude
loading (VAL) rather than constant amplitude loading (CAL).
APPROACHES TO LIFE PREDICTION
Life prediction approaches that are based on the presence of a crack are termed crack
growth methods, while those based on damage assessment during initiation (formation
of a crack a few millimetres in size) are termed crack initiation methods (Ringsberg &
Lindback 2001). The total life to failure consists of a crack initiation phase followed by
crack propagation. While the initiation period can be modelled by surface stress or strain,

crack propagation requires consideration of stress intensity factors and re-meshing of the
cracked region. Thus, fatigue failure is defined as the time to initiate a crack. Since most
of the time to failure for smooth components is spent in crack initiation (Bonamy 2009),
it can be argued that this is an appropriate and also a conservative failure criterion.
The crack initiation approach involves two operations in converting the load history,
component geometry, and material input into the predicted life. These operations must be
performed sequentially, as shown in Fig. 1. A study by Ridnour (2003) developed models
that have been used to predict the fatigue life of the trailer surge brake. The fatigue life
prediction is divided into two parts: initiation and propagation. The fatigue crack
initiation life prediction uses a multiaxial local strain approach, and failure is assumed to
occur when the crack has reached
2 mm in length. The fatigue crack propagation life prediction uses the FLAGRO software
for which this software was developed by NASA. The results showed that the simulation
can be modified to represent of the tested vehicle also. In addition, the fatigue life and
durability of the vehicle can also be predicted with a model and data obtained from some
sensors which have been placed on the vehicle components.
Fig.1. The information path in a crack initiation life prediction
Material Component Loading
Properties Geometry History

Stress-Strain
Analysis
Damage
Analysis
Fatigue
Life
Firstly, stresses and strains at the critical site are estimated. Then, the critical location or
local stresses and strains are used to compute damage that is algebraically added up
throughout the history until a critical damage sum (failure criterion) is reached. The point
in the history at which the failure criterion is met is the predicted life (Darrell 1968).
Some methods implement the crack initiation approach by computing damage from
remote loads or stresses that are simply scaled to account for the notch. These methods
that calculate damage using remote or nominal stresses and strains are sometimes
called nominal methods. Other methods implement the crack initiation approach by
calculating stresses and strains at the notch, so that damage is assessed directly in terms
of local strains and stresses. These methods that utilise local stresses and strains are often

called local methods or critical location methods. The main analysis methods available
are Stress-Life, Strain-Life and Crack-propagation.
Low-cycle strain-controlled fatigue behaviour became promoted with the Coffin (1954)
and Manson (1965) relationship between plastic strain amplitude and fatigue life. In
addition, the strain-based approach to fatigue problems is widely used for correlating
with low-cycle fatigue too, as reported in Hurley (2008) and Koh (2009). When the load
history contains large overloads, significant plastic deformation can exist, particularly at
stress concentrations, and load sequence effects can be significant. In these cases, the
strain-life approach is generally superior to the stress-life approach for cumulative fatigue
damage analysis (Ralph et al., 2000).
The most common application of the strain-based approach, however, is in fatigue of

notched members. Since the fatigue damage is assessed directly in terms of local strain,
this approach is called the local strain approach too. Thus, it is common that the service
loadings caused by machines and vehicles is evaluated using a strain-life fatigue damage
approach (Refngah et al., 2009).
Strain-life fatigue curves plotted on log-log scale are schematically shown in Fig. 2,
where Nf or 2Nf is the number of cycles or reversals to failure, respectively. The strain-
life curve has been resolved into the elastic and the plastic strain components from the
steady-state hysteresis loops. At a given life, Nf , the total strain is the sum of the elastic
and plastic strains. Both the elastic and plastic curves can be approximated as straight
lines. At large strains or short lives, the plastic strain component is predominant, and at
small strains or longer lives the elastic strain component is predominant. The strain-life
method has achieved the status of industry standard in the North American automotive
industry (Conle & Chu, 1997) while in Europe, the automotive industry shows a
preference for stress-life methods (Berger et al., 2002).
Fig.2. Strain-life curves with elastic and plastic strain components (Stephens et al., 2001)
f
Strain amplitude (log scale)
c
1 Total
f/b
b
Elastic 1
Plastic
Transition life 2NT 2Nf

Reversal to failure (log scale)

The service loads of components of machines, vehicles, and structures are typically
analysed for fatigue life using crack growth approaches. This approach is suitable for
high capital value items such as large aircraft, the space shuttle, pressure vessels and oil
rigs (Dowling 1999). The ability to inspect for cracks and monitor their growth until a
maximum allowable defect size is reached normally enables the useful life to be extended
beyond the original design safe life. However, it is not generally feasible to perform crack
inspection for inexpensive components that are made in large numbers.
Fatigue life prediction represents one of the applications for finite element. Among the
plastic strain models which are widely used are the Coffin-Manson, Morrow and Smith-
Watson-Topper (SWT) models. It is indicated by the straight-line curves and the size of
the hysteresis loop in Fig. 2. The intercepts of the two straight lines at 2Nf =1 are
f / E for the elastic component and f for the plastic component. The slopes of the
elastic and plastic lines are b and c, respectively. This provides Eq. 1 for strain-life data
(Coffin 1954; Manson 1965):

a e

p

f '
2 N f
b
f ' 2 N f
c (1)
2 2 2 E
where / 2 is the total strain amplitude, a , e / 2 is the elastic strain amplitude

/ 2E a / E , p / 2 is the plastic strain amplitude / 2 e /2 ,
is the fatigue ductility coefficient, c is the fatigue ductility exponent, is the
f f
fatigue strength coefficient, b is the fatigue strength exponent, E is the modulus of

elasticity, and / 2 is the stress amplitude.
One method, often referred as the (Morrow mean-stress correction relations) replaces

f with f m in Eq. 1, where m is the mean stress (Morrow 1968), such that:
'
a
f m
2 N f
b
f ' 2 N f c
2 E (2)
In Eq. 2, m
is taken to be positive for tensile values and negative for compressive
values. This equation predicts the tensile mean stress is detrimental and the compressive
mean stress is beneficial. Eq. 2 predicts more effect of the mean stress relationship which
can be derived at long lives. Alternative version of the Morrow mean-stress were both the
elastic and plastic terms are affected by the equivalent mean stress, which is
mathematically given by Manson & Halford (1981)
c
' f ' b
a
f m
2 N f
b
f '
'
m

2 N f
c (3)
2 E f

Another the strain-life mean stress correction model was suggested by Smith et al.
(1970), or often called the SWT parameter. This relationship was based on strain-life test
data which was obtained at various mean stresses. Thus, the SWT expression is
mathematically defined as
a E ' 2 N ' E 2 N
2b bc
'
2
max f f f f f
(4)
where ( m m a and a is the alternating strain). This equation is based on the

assumption that for different combinations of strain amplitude, a and mean stress, m ,

max
a remains constant for a given life. The main task performed during
the product
durability analysis is the fatigue life assessment of components such as engine parts,
suspension parts and body structures (Haiba et al., 2003).
MATERIAL CHARACTERISATION
Chemical Composition Test

The purpose of analysing the chemical composition of a lower suspension arm sample is
to enable classification. One sample was cut from the lower suspension arm using a
cutter. The sample was subsequently grounded with successive SiC papers (grit size of
200-1200) and then polished with polishing cloth and Alumina solution with the grain
size of 6 m, and finally 1 m. The measured chemical values as shown in Appendix 1,
were acquired using an Arc Spark Spectrometer machine test.
Hardness Test
The Rockwell hardness tester (SHIMADZU model) has been used to determine hardness
Rockwell (HRC). Fig. 3 shows hardness tests samples for the lower suspension arm
material (AISI 1330H steel) and the experimental tests material (SAE1045 steel).
Fig.3. The hardness test samples for: (a) SAE 1045 steel, (b) AISI 1330H steel
(a) (b)
Material Selection
From the chemical composition test result of Appendix 1, it was found that the material
used for the lower suspension arm is manganese alloy steel. There are additional elements

that have been added by the company to the standard material (ASM Specialty Handbook
1996) for the purpose of component properties improvement. Thus, it is difficult to obtain
the exact material to perform the experimental tests. Therefore, another material
represented by the SAE1045 steel has been chosen for the experiment and for the FEA
optimisation purposes, which is often used for the suspension components of
automobiles. Suspension system arms can be fabricated using cast steel (Mahishi 2005).
The results of the chemical composition test for the SAE1045 steel is shown in
Appendix 2.
Specimens Fabrication
The material was in the form of round cross section bar with 13mm diameter made of
SAE1045 steel. Identical bar specimens were prepared for the monotonic and fatigue test.
The photographs of tensile and cyclic specimens are shown in Fig.4. The specimens were
subsequently grounded with successive SiC papers (grit size of 1202000) and
mechanically polished. The mean roughness of samples (Ra) was 0.1
cracks begin to occur at a free surface, the condition of the fatigue specimen surface is
important in fatigue life test (Dowling 1999).
Fig. 4. The photograph of specimens used for: (a) tensile, (b) cyclic test
(a)
(b)
All specimens were cut using commercial turning machine and fabricated using CNC
milling machine (CINCINNATI MILACRON of ACRAMATIC, 850 SX-IDS CNC
CONTROLS). The specimen gauge section were polished using 120, 400, 600, 800,
1000, 1200, 1800, 2000 grades of sand paper. The polished surfaces were carefully
examined to ensure complete removal of machining marks in the test section (Dowling
1999).
Geometry and Finite Element Model

An automotive lower suspension arm component for a 2000 cc Sedan car, as shown in
Fig.5., was used for this study. The three-dimensional model was drawn using Computer

Aided Design (CAD) software package for the engineering design. The produced file
(*.model format) has been read by FE software for modelling and simulating purposes.
Meshing is performed to discretise the geometry of the lower suspension arm created into
small pieces called elements using a set of nodes, as shown in Fig.6.
Fig. 5. An automotive lower suspension arm component
Fig. 6. FE model of automotive lower suspension arm
RESULTS AND DISCUTION
The results that have been obtained in both of the experimental and analytical FEA stages
have been discussed. The component critical location was validated using experimental
road data collection. The, component fatigue life was predicted based on the pseudo-
static method.
Experimental Determination of Material Properties
Material properties represented by chemical composition, hardness, monotonic and cyclic

properties have been determined.
Chemical Composition
According to Appendix 1, the steel sample can be classified as manganese alloy steel. It
is AISI 1330H steel since a carbon content range 0.27-0.33%, a manganese content range
1.4-2%, sulfur <= 0.04, silicon range 0.15-0.35%, and phosphorous <= 0.035% (ASM
Specialty Handbook 1996). It represents the fabricated material for the 2000 cc sedan
lower suspension arm and was the material used in simulation work. The usage of the
SAE 1045 steel in the experimental tests was for the optimisation purposes. The chemical
composition of the SAE 1045 steel received material is shown in the Appendix 2.
Material Hardness
Table 1 shows the measured and standard Rockwell hardness values for both
AISI 1330H steel and SAE1045 steel (ASM Specialty Handbook 1996). It was noticed

that the hardness of AISI 1330H steel is higher than SAE1045 steel. This is because the
carbon percentage in SAE1045 steel (0.43%) is higher than in AISI 1330H steel (0.30%)
while the manganese percentage in SAE1045 steel (0.69%) which is less than in AISI
1330H steel (1.43%). Carbon element controls the hardness and it also has a significant
effect on the hardenability. However, the manganese element is one of the most
commonly used element for the promotion of hardenability. As the hardness increases,
damage accumulation rate increases (Tehrani and Saket 2009), the longer life could be
for SAE1045 steel because the hardness value is lower than AISI 1330H steel.
Table 1. The measured and standard HRC for AISI 1330H and SAE 1045 steel
Hardness AISI 1330H steel SAE 1045 steel
Measured HRC 34 26
Standard HRC 25 13
Monotonic and Cyclic Properties

The experimentally determined properties from Fig.7. that shown in Table 2 are: modulus
of elasticity, yield stress, ultimate tensile strength, percentage of elongation, fracture
strength and ultimate tensile strength. Further details of experimental work that was
performed to obtain the SAE1045 steel monotonic and cyclic properties can be found in a
previous study by the authors (Kadhim et al., 2012).
Fig. 7. Engineering and true stress-strain curve for SAE1045 steel

1000
900
800
700
Stress (MPa)
600
500
400
300
200
100
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Strain (%)
Engineering stress-strain True stress-strain
Table 2. Monotonic mechanical properties of SAE1045 steel (Kadhim et al., 2012)

Parameters Values
Ultimate tensile strength, 857
Modulus of elasticity, E (MPa) 207
Fracture strength, 730
Percent elongation, 7
Static yield stress 0.2%, 652

The total strain values, which were used in the CAL fatigue tests, have been calculated
using the well-known Ramberg-Osgood relationship. The MSC. Fatigue software was
used to calculate the material constant (K = 1412) and the strain hardening exponent (n =
0.15) for the SAE1045 steel. These are calculated according to the inputs represented by
monotonic properties, as in Table 2. Substituting the modulus of elasticity value E =
207000 MPa, K and n into Ramberg-Osgood relationship provided the complete
Ramberg-Osgood relationship for this material as the following:
1
e p 0 . 15
(5)

207000 1412
In accordance to ASTM E739 (1998), the true plastic strain amplitude ( p /2) is the
independent variable and the true stress amplitude ( /2) is the dependent variable.
The true plastic strain amplitude was calculated based on Eq. 6 as follows
p
(6)
2 2 2E
The Coffin-Manson relationship parameters for the strain-life curves are calculated based
on both the elastic and plastic curves that can be approximated as straight lines. These
parameters are presented in Table 3.
Table 3. Cyclic properties by means of the Coffin-Manson relationship for the SAE1045
steel (Kadhim et al., 2012)
Parameters Values
Fatigue strength coefficient, 1193
Fatigue strength exponent, b -0.074
Fatigue ductility coefficient, 0.33
Fatigue ductility exponent, c -0.49
Substituting the Coffin-Manson coefficients into Eq.s 1, 2 and 3 provided the Coffin-
Manson, Morrow and SWT models, respectively, as the following expressions:
Coffin-Manson: a
1193
2 N f
0 . 074
0 . 33 2 N f
0 . 49 (7)
207000
1193 m
Morrow:
a 2 N f
0 . 074
0 . 33 2 N f 0 . 49 (8)
E
SWT: max a E 1423249 2 N f

0 . 148
121005990 2 N f
0 . 56
(9)
Characterising Materials Properties using Finite Element Analysis

In the design where cyclic loading is involved, considering monotonic strength as one of
the design variable is not sufficient to obtain a safe and reliable design. This is due to the

existence of the cyclic softening or hardening (Dowling, 1999). Therefore, it is essential
to consider cyclic deformation of the component material since the component is
subjected to the cyclic loading.
The fatigue ductility coefficient and exponent are the material properties used to describe
the N behaviour to evaluate the low-cycle fatigue performance. The lower
suspension arm is fabricated from the AISI 1330H steel as explained earlier, and will be
optimised to SAE 1045 steel. The monotonic and the cyclic properties of both materials
will be discussed in the subsequent section.
Comparing Monotonic Deformation Behaviour

A summary of the monotonic properties for both materials is provided in Table 4. Fig. 8.
presents a direct comparison of the monotonic stress-strain behaviour for both materials.
The E value of the SAE 1045 steel in the standard was found to be at 204 MPa, which is
somewhat less than what was measured in the experimental. This could be due to the
characteristics of products. In order to investigate the difference, further metallurgical
and microstructural analyses are required. However, this is not within the main scope of
this study.
Table 4. Summary of the mechanical properties and their comparative

SAE 1045 steel AISI 1330H steel
Monotonic Properties
Modulus of elasticity, E, GPa 207 200
Yield strength (0.2% offset), S y , MPa 652 1034
Ultimate tensile strength, Su , MPa 857 1158
Cyclic and Fatigue Properties

Fatigue strength coefficient, f , MPa 1193 1737
Fatigue strength exponent, b -0.074 -0.087

f 0.49 0.38
Fatigue ductility coefficient,
Fatigue ductility exponent, c -0.33 -0.58
Cyclic strength coefficient,K , MPa 1412 1911
Cyclic strain hardening exponent, n 0.15 0.15
The ultimate tensile strength of the AISI 1330H steel was found to be higher 25% of SAE
1045 steel. Moreover, the yield strength of SAE 1045 steel is lower 37% of the AISI
1330H steel. The fatigue strength of AISI 1330H steel was found to be higher 26% of
SAE 1045 steel. Based on the strength, the AISI 1330H steel is superior to another
material. Although the comparison of the monotonic strength data as one of the major
design parameters shows the superiority of the AISI 1330H steel, it is not sufficient to

consider only the monotonic properties as the design parameters when cyclic loading is
involve.
Comparing Cyclic Deformation Behaviour

The cyclic stress-strain curve reflects the resistance of a material to cyclic deformation.
a- Steady state cyclic deformation

The stress-strain curves of both materials are shown in Fig.8. It could be observed that
the Ramberg-Osgood Equation provides a good representation of cyclic deformation
behaviour. It also shows that the AISI 1330H steel has higher strength with its yield point
well above that those of the SAE 1045 steel. It is concluded that the SAE 1045 steel
perform better in the monotonic condition in the lower strain range while AISI 1330H
steel performed better in the cyclic condition in the higher strain range.
Fig. 8. Monotonic stress-strain curves for SAE 1045 steel and AISI 1330H steel
b- Strain-controlled fatigue behaviour and comparisons

The constant amplitude strain-controlled fatigue data were used to determine the strain-
life curve. Eq.s 10 and 11 relate the true strain amplitude to the fatigue life.
e p

2 2 2 (10)
f
2 N
f
b
f 2 N f c
E (11)
where f is the fatigue strength coefficient, b is the fatigue strength exponent, f is the
fatigue ductility coefficient, c is the fatigue ductility exponent, E is the monotonic

modulus of elasticity, and 2 N f
is the number of reversals to failure (which was
defined at 50% load drop, as recommendation by ASTM E606 (1998). A summary of the
cyclic properties for the both materials is provided in Table 4.
Fig.9. represents the two materials strain-life behaviour under CAL. This figure shows
that the SAE 1045 steel at short life or high strain exhibits better fatigue resistance. In the
lower suspension arm design, cyclic behaviour is a major concern when designing
component that is subjected to occasional overloads, particularly for notch components,
where significant plastic deformation can occur. It is also observed that the SAE 1045
steel and the AISI 1330H steel strain-life curves cross each other which means share the
same life in a certain strain value.
The comparison of the strain-life fatigue behaviour for these two materials demonstrate
the superiority of the SAE 1045 steel with respect to the low cycle fatigue. Based on the
comparison between the two materials strain-life, conclusion were made that it is difficult
to predict which steel would be more durable based on the curve. Significantly, it reflects
the need for further fatigue analysis under VAL. since several studies showed a
significant difference between CAL and VAL fatigue test results based experimental
investigations (Yan et al., 1992; Agerskov, 2000; Lee et al., 2007; Alaoui et al., 2009;
Carvalho et al., 2010)
Fig. 9. Strain-life curves for SAE 1045 steel and AISI 1330H steel
Fatigue Life Assessment under Variable Amplitude Loading

For the purpose of fatigue lives assessment for the three types of road data signals,
experiences have been gained by conducting VAL tests. An INSTRON 8801 closed-loop
servo-hydraulic axial load frame was used to conduct the test. Further details about the
VAL tests are available in a previous study by Kadhim et al. (2012).

In addition, fatigue lives of the three road types were predicted based FEA. A three-
dimensional (3D) model of the lower suspension arm was developed using the CATIA
software. Then a 3D FE model was adopted for the simulation, as shown in Fig.10. with
the applied load and constraints.
Fig. 10. Three-dimensional finite element model
The linear static analysis was performed because of its simplicity (Lakshminarayana
2004) using the MSC.NASTRAN FE software in order to determine the stress and
strain results from the FE model. From the acquired result values, the maximum principal
stress of 293 MPa and strain of 1.34 x 10-3 occurring at node 447 were obtained. These
values were validated through a comparison with the collected road strain data values.
The data has been experimentally collected using the specific data acquisition system by
fixing the strain gauges at different positions as explained in a previous study of the
authors (Kadhim et al., 2011).
The time domain fatigue signals were edited for shortening the signal length with the
removal of low amplitude cycles which have no effect on total damage as approved in a
study by Kadhim et al. (2011). In this case, the removed data was assumed as non-
damage segments because the departures of those segments did not change the value of
the total fatigue damage.
Figs. 11~13 show three of road time series before and after editing which were used to
predict component fatigue life. The country road signal in Fig.11., shows the existence of
several bumps that is related to fatigue damage event. In the Pavement data (Fig.12.),
only one second of 75 seconds was omitted based on the FDE technique. The FDE
technique was able to detect only one clear sudden change that occurred in the fatigue
data (as indicated in the green highlighted area) because of the road nature, the signal
amplitude values exhibits the same trends. Finally, Fig.13. shows that the detection of
sudden changes by using FDE technique tends to be less than the country road and more
than the pavement road. In general, there are differences between the high and low
amplitudes of strain data which was observed for each road type. The country road has
more variety in the strain amplitude range than the highway followed by the pavement.
The signals pattern differences were the main reasons for the data editing limitation for
each road type. At the same time, the setting of the retained fatigue damage value was set
on 100 % for all the road signals.

Fig. 11. Country road strain time signal: (a) signal editing process, (b) the 205 seconds
edited signal
a)
b)
Time (Seconds)
Fig. 12. Pavement strain time signal: (a) signal editing process (b) the 74 seconds edited
signal
a)
b)
Time (Seconds)

Fig. 13. Highway strain time signal: (a) signal editing process (b) the 122 seconds edited
signal
a)
b)
Time (Seconds)
The most commonly used statistical parameters are the root-mean-square (R.M.S.) value
and the kurtosis (Hinton, 1995). The R.M.S. value, which is the 2nd statistical moment,
represents the vibration signal energy in a time series. It is used to quantify the overall
energy content of the signal and is defined by the following equation:
1 2
1 n
R.M.S.
n

j 1
x j
2

(12)
The kurtosis, which is the signal 4th statistical moment, represents the amplitude range in
a time series. It is a global signal statistic which is highly sensitive to the spikiness of the
data.
x
n
1 4
K j x
n ( R.M.S. )4 j 1
(13)
For the country road, the lengths of the original and edited signals were 385 and 205
seconds, respectively or 53.2% of the original signal, as illustrated in Fig.11. and Table 5.
Refering to the same table, the number of points of the original and edited signals were
192800 and 94740 points, respectively, or 49.1% of the original signal. The R.M.S. and
the kurtosis difference between the editing and the original signals were 24% and
16.91%, respectively.
The pavement edited signal has the time length of 74 seconds or 98.6% of the original
signal, as illustrated in Fig.12. and Table 5. Referring to the same table, the number of
points of the original and edited signals were 37502 and 37001 points, respectively, or
98.6% of the original signal. This time length was obtained when the R.M.S. and kurtosis
differences of the edited signal were less than 10% of the original signal.

Finally, the edited highway road signal length was 122 seconds or 90.3% of the original
signal, as illustrated in Fig.13. Referring to Table 5, the number of points of the original
and edited signals were 67502 and 60490 points, respectively, or 89.6% of the original
signal. The R.M.S. and the kurtosis difference between the editing and the original
signals were 1.8% and 4.27%, respectively.
The R.M.S. value shows the overall energy content for the data signal. Therefore for
higher damage values the R.M.S. values are theoretically higher. The same occurrence
was found for the three types of road signal as shown in Table 5. The R.M.S. values for
the country road original and edited signals are 85 and
106 , respectively that were higher than the value of pavement road original and edited
signals at the value of 62 and 63 respectively. It was then followed by the value for
the highway road original and edited signals at 53 and 54 respectively. The second
statistical parameter represented by kurtosis value as shown in the same table shows the
same trend of the R.M.S. value for each signal with some variation in the country road
edited signal. The variations in kurtosis values are due to the randomness of the data and
the variety in the size of each segment.
Table 5 The characteristics of signal length and number of points with global statistics
between original and edited signals
Signal Signal statistical properties
type
Kurtosis difference (%)

Signal length (seconds)
No. of points ratio (%)
R.M.S difference (%)

(%)
No. of points
R.M.S (
Time ratio
Kurtosis
Country 385 53.2 192800 49.1 85 24 4.61 16.91

original
Country 205 94740 106 3.83
edited
Pavement 75 98.6 37502 98.6 62 1.6 4.50 0.88
original
Pavement 74 37001 63 4.46
edited
Highway 135 90.3 67502 89.6 53 1.8 3.98 4.27
original
Highway 122 60490 54 4.15
edited
Note: Ratio (%) = (editing / original) x 100%
Statistical difference = (statisticsediting statisticsoriginal)/statisticsoriginal x 100%

The experimental fatigue life result was compared with SWT fatigue life prediction with
72, 60, and 55% for the country road, pavement, and highway, respectively. For loading
sequences, which are predominantly tensile in nature the SWT approach is more
conservative; therefore, it has been selected for the current road signals fatigue
assessment. Based on the uniaxial VAL fatigue test data results in Table 6, it was noticed
that the shortest life is for the country road signal followed by pavement and the highway
road signals for the experiment and FEA prediction. Thus, the longest life was found for
the smoothest surface which represented by highway road. In addition, the results of
SWT mean stress correction method show the same life for both the original and the
editing signals, this is due to remove only the signals that no damage affects on the
original signal.
Table 6. Experimental fatigue lives and differences with SWT model for three different
road data signals
Signal type Experimental SWT x103 Fatigue life difference
4
x10 for original between experiment and
and edited SWT (%)
signals
Country road 1.79 6.26 72
Pavement 3.90 9.02 60
Highway 9.25 15.35 55
CONCLUSIONS
Experimental and computational durability assessment approaches of the lower

suspension arm under actual service loadings were presented. The FEA results can be
considered as an initial process for any component design unless it was in combination
with the experimental tests to minimize the number of physical prototypes.
The results show a big difference between experimental and FEA fatigue estimation life.
In a nutshell, the results reflect the need for new models associated with FEA package
which can be more suitable to predict fatigue life under variable amplitude loading
(VAL) rather than constant amplitude loading (CAL).
ACKNOWLEDGEMENTS
The financial support (UKM-OUP-TKP-23-188/2011) from the Institute of Space

Science (ANGKASA) of the Universiti Kebangsaan Malaysia is gratefully
acknowledged.
REFERENCES
Agerskov, H. (2000). Fatigue in steel structures under random loading. Journal of

Constructional Steel Research 53: 283305.

Alaoui, A.E., Thevenet, D. and Zeghloul, A. (2009). Short surface fatigue cracks growth
under constant and variable amplitude loading. Engineering Fracture Mechanics 76:
23592370.
ASM Specialty Handbook. (1996). Carbon and Alloy Steels, edited by J. R. Davis and
Associates, ASM International, Metals Park, OH, Metals Handbook, Vol. 1 properties
and Selection: irons, Steels, and High-Performance Alloys, ASM International 10th Ed.
ASTM Standard E606-92. (2004). Standard practice for strain-controlled fatigue

testing. American Society for Testing and Materials.
Berger, C., Eulitz, K.-G., Heuler, K.-L., Kotte, H., Naundorf, W., Schuetz, C., Sonsino,
A., & H. Zenner. (2002). Betriebsfestigkeit in Germany an overview. International
Journal of Fatigue 24(6): 603-625.
Bonamy, D. (2009). Intermittency and roughening in the failure of brittle heterogeneous

materials. Journal of physics D: applied physics 42: 1-21.
Carvalhoa, A.L, Martinsb, J.P. and Voorlwadc, H.J. (2010). Fatigue damage
accumulation in aluminum 7050-T7451 alloy subjected to block programs loading under
step-down sequence. Procedia Engineering 2(1): 20372043.
Coffin, L. (1954). A study of the effect of cyclic thermal stresses on a ductile metal.
Transactions of ASME 79: 931-950.
Conle, A. & Chu, C.C. (1997). Fatigue analysis and the local stress-strain approach in
complex vehicular structures. International Journal of Fatigue 19(1): S317-S323.
Darrell, S., Harold, R., Stephen, D., Steve, T., Brian, L. & Drew, N. (1968). Fatigue
Design Handbook, Advances in Engineering. Warrendale: Society of Automotive
Engineers.
Devlukia, J. and Bargmann, H. (1997). Fatigue assessment of an automotive suspension

component using deterministic and probabilistic approaches. 2nd International
Symposium of Fatigue Design, pp. 436-445.
Dowling, N. E. (1999). Mechanical Behaviour of Materials: Engineering Methods for

Deformation, Fracture and Fatigue. New Jersey: Prentice Hall.
Erjavec, J. (2006). TechOne: Automotive Suspension and Steering. USA: Thomson

Delmar Learning.
Haiba, M., Barton, D., Brooks, P. and Levesley, M. (2002). Review of life assessment
techniques applied to dynamically loaded automotive components. Computer Structure
80: 481-494.

Haiba, M., Barton, D., Brooks, P. and Levesley, M. (2003). The development of an
optimisation algorithm based on fatigue life. International Journal of Fatigue 25: 299-
310.
Hinton, P. R. (1995). Statistics Explained: A Guide for Social Science Students.

Routledge, London.
Hurley, P., Whittaker, M., Williams, S. & Evans, W. (2008). Prediction of fatigue
initiation lives in notched Ti 6246 specimens. International Journal of Fatigue 30(4):
623-634.
Kadhim, N.A., Abdullah, S. and Ariffin, A.K., (2011). Effect of the fatigue data editing
technique associated with finite element analysis on the component fatigue design
period, Materials and Design, Elsevier Ltd, Vol 32, No 2, pp 10201030.
Kadhim, N.A., Abdullah, S. and Ariffin, A.K., (2012). Effective strain damage model
associated with finite element modelling and experimental validation, International
Journal of Fatigue, Elsevier Ltd, Vol 36, No 1, pp 194205.
Kim, J., Lei, L. and Kang, B. (2003). Preform design in hydroforming of automobile
lower arm by FEM. Journal of Process Technology (138): 58-62.
Koh, S. K. (2009). Fatigue analysis of an automotive steering link. Engineering Failure

Analysis 16: 914922.
Lee, Y., Tjhung, T. and Jordan, A. (2007). A life prediction model for welded joints
under multiaxial variable amplitude loading histories. International Journal of Fatigue
29: 11621173.
Mahishi, J.M., (2005). Nonlinear static and multi-axial fatigue analysis of automotive
lower control arm using NEINASTRAN. http://www.nenastran.com/
newnoran/conferencePaper2/10_CPNonlinearStaticMultixialFatigueAnalysisAutomotive
LowerControlArmUsingNEiNastran.pdf
Manson, S. S. (1965). Fatigue: a complex subject some simple approximation.

Experimental Mechanics 5: 193-226.
Manson, S. & Halford, G. (1981). Practical implementation of the double linear damage
rule and damage curve approach for treating cumulative fatigue damage. International
Journal of Fracture 17: 169-192.
Morrow, J. D. (1968). Fatigue Properties of Metal, Fatigue Design Handbook, Society

of Automotive Engineers.

Mrzyglod, M. and Zielinski, A. (2007). Parametric structural optimization with respect
to the multiaxial high-cycle fatigue criterion. Structural Multidisciplinary Optimization
33: 161-171.
Ralph, I.S., A. Fatemi, Robert, R. & Henry, O. (2000). Metal Fatigue in Engineering.
New York: John Wiley and Sons Inc.
Refngah, F.N., Abdullah, S., Jalar, A. & Chua, L.B. (2009). Life Assessment of a
Parabolic Spring under Cyclic Strain Loading. European Journal of Scientific Research
28(3): 351-363.
Ringsberg, J. & Lindback, T. (2001). Rolling contact fatigue analysis of rails including
numerical simulations of the rail manufacturing process and repeated wheel-rail contact
loads. International Journal of Fatigue 25: 547-558.
Smith, K., Watson, P. & T. Topper. (1970). A stress-strain functions for the fatigue of
metals. Journal of materials 5: 767-767.
Song, X. J., Son, H. J., Park, J. H., Lee, K. H. and Park, Y. C. (2009). Metamodel-based
optimization of a control arm considering strength and durability performance.
Computers and Mathematics with Applications 60 (4): 976-980.
Stephens, R.I., Fatemi, A., Stephens, R.R. & Fuchs O.H. (2001). Metal fatigue in
engineering. USA: John Wiley & Sons.
Tehrani, M.P H. and Saket. (2009). Fatigue crack initiation life prediction of railroad.
7th International Conference on Modern Practice in Stress and Vibration Analysis. UK.
Yan, X., Cordes, T.S., Vogel, J.H. and Didinger, P.M. (1992). A property fitting
approach for improved estimates of small cycle fatigue damage. USA: Society of
Automotive Engineers.
APPENDIXES
1- Material Chemical Composition for AISI 1330H Steel Specimen

2- Material Chemical Composition for SAE 1045 Steel Specimen

Mechanical Tolerance Synthesis on theMultimedia
Oriented Systems Transport Standard Transceiver
Module
Michael Vervaeke, Els Moens, Youri Meuret, Heidi Ottevaere,Carl Van
Buggenhout, Piet De Pauw and Hugo Thienpont
INTRODUCTION
The advent of Plastic Optical Fibre (POF) opened perspectives for numerous ap-
plications in the field of datacommunications. POF is increasingly popular in
the automotive industry as a robust, lightweight, electromagnetic interference free
and low-cost alternative to electrical wiring for high-speed entertainment, naviga-
tion and data acquisition systems in cars. Moreover, POF is attributed to be easy
and cheap to install(Daum et al., 2002) . The tight specifications for the working
conditions in automotive applications set the main challenge for the introduction
of datacommunication systems based on POF in the automotive sector since these
systems are required to remain fully functional in a temperature range from -40 C
to 115 C .
We designed in collaboration with Melexis a misalignment-tolerant optical
coupling system (Vervaeke et al., 2009) according to the Media Oriented Systems
Transport standard (Grzemba, 2011), see Fig. 1 , to convey the divergent beam
from a Resonant Cavity Light Emitting Diode (RCLED) into a Step-Index (SI)
multimode POF mounted in a detachable ferrule. In this contribution we describe
the methodology to synthesize the dimensions and tolerances on the optical com-
ponents in the coupling system taking into account the geometrical and dimen-
sional tolerances on the mechanical mounts of the optical parts. A Monte Carlo
optimisation algorithm on the full three-dimensional (3D) description of the com-
plete RCLED package and detachable POF ferrule was used to allow a realistic
modelling of all misalignments that could occur in the production chain. Besides
pinpointing the crucial tolerance sets in production we were able to select the best
suited system according to manufacturing and assembly capabilities.
Currently the transceiver module, based on the MOST 150 standard, is avail-
able on the market and taken up by BMW, Daimler and Audi in high-end car

Figure 1: The MOST 150 SMD transceiver and POF fiber ferrules. Courtesey
Melexis.
models.
DESCRIPTION OF THE OPTICAL PHYSICAL LAYER OF MOST
The MOST network architecture is a logical ring, usually implemented as a phys-

ical ring, consisting of POF. Fig. 2 shows the basic principle of optical fibers, and
multimode SI POF more in particular. An optical fiber consists of a low index
refractive material, here an ncore of 1.41 with a diameter of 980 m, surrounded
by a high index cladding, in this case with a refractive index of ncladding of 1.49
and a diameter of 1000 m. Light inside this fiber bounces off the interface be-
tween core and cladding travelling along a zigzag path from entrance facet to
output facet. The maximum angle at which light rays can bounce off the interface
between the low and high refractive index materials depends only on the mate-
rial refractive indices. This maximum angle limits the angle at which light can be
launched inside the POF and similarly limits the cone of light emitted at the output
facet. A common measure to characterize this acceptance angle is the Numerical
Aperture (NA) which relates to half the total cone angle or acceptance angle as:

NA = sin() = n2core n2cladding (1)
The POF used in the MOST standard has an NA of 0.48 .
In comparison other fiber types such as telecom glass fibers have NAs down to
0.12 for single mode high-speed applications with a cladding diameter of 120 m

Figure 2: Principle of light propagation in optical fibers(Grzemba, 2011).
and a core of 9 m. Similarly other POF types such as step index POF with NA
of 0.3 , gradient index POF, smaller core diameter or tapered end POF, were not
considered since they are not used or even considered today in automotive appli-
cations due to too high bending losses, unavailability of transceiver modules or
low cost connectors, or not compatible with low cost installations and the difficul-
ties in coupling light into the small acceptance angle. However the methodology
used in this study is fully applicable to these cases as well and is in principle not
even limited to guided wave approaches.
While high NA POF such as the one used in the MOST standard deliver easy
coupling, they exhibit high modal dispersion due to the large difference in pathway
between rays entering at low and high angles. This modal dispersion leads to pulse
broadening and hence a lower bandwidth-length product in comparison with low
NA fibers. In case of the MOST POF 320 ps of pulse broadening per meter fiber
length can be expected, resulting in a maximum datarate of 150 MBit/s for the
longest 20 m link lengths expected in automotive applications. In reality a source
will not excite all ray angles equally resulting in higher datarates going up to
500 MBit/s for RCLEDs.
The polymethyl methacrylate (PMMA) material that makes up the POF re-
sults in an attenuation of 0.14 dB/m at the wavelength MOST devices transmit
(650 nm). Due to the low resistance to thermal peaks manufacturers revert to other
materials such as polycarbonate for fibers in environments with elevated temper-
atures. These fibers have a minimum attenuation of 0.4 dB/m at the earlier given
wavelength. Another source of losses are sharp bends. It was shown that up to
25 mm bending radius no considerable extra loss could be measured. At 10 mm
the excess attenuation rises to 0.5 dB on top of the normal attenuation.
The POF is protected by a black protective cladding buffer which is perma-
nently fixed to the optical fiber. A cable coating further protects the POF from
mechanical damage and other influences such as fluids and humidity. The black
buffer itself is an attachment interface for the fiber termination, which consists of
either a pressed metallic ferrule or an ultrasonic or laser welded plastic ferrule as
shown in Fig. 3 . The ferrule has a specific geometry to allow accurate mating in

Figure 3: Termination technologies for polymer optical fibers. From top to bot-
tom: crimping of a metallic ferrule, ulstrasonic welding and laser wlding of plastic
ferrules (Grzemba, 2011).
connector sleeves or transceiver package receptacles. We have implemented the

3D geometry of the ferrule mating interface and POF together with the transceiver
receptacle in a polygon model for tolerance synthesis as will be explained in fur-
ther sections.
The standard light source in the MOST description is a light emitting diode
(LED) emitting at a wavelength of 650 nm. This emitter is a very cost-efficient
component for low-cost PMMA POF networks with bandwidths below 100 MHz.
Using suitable technologies the efficiency of the LEDs can be driven up to 20 %
and with specialty driver circuits the modulation bandwidth can be extended be-
yond 100 MHz.
RCLEDs exhibit a much larger modulation bandwidth,
beyond 500 MHz and are now considered in the MOST 150 standard as a light
source. The RCLED has a circular emitting surface with a diameter of 83 m .
Due to mechanical constraints in the package design of the RCLEDs, the input
facet of the POF has to be at a distance of 443 m from the source. The minimum
power to be coupled into the POF is set to -3.5 dBm for all operating conditions in
order to satisfy the power budget margins in the MOST standard.
The receiver side of the MOST links are implemented with Silicon-PIN photo-
diodes due to their cost-efficiency. The receivers in the MOST 150 SMD package
are large-area photodiodes to match the large core of the PMMA SI POF, lead-
ing to a bandwidth reduction to 100 MHz but a fairly good receiver sensitivity of
-30 dBm and a responsivity of 0.47 A/W at the chosen wavelength of 650 nm.
RCLED, SI PIN photodiode, driver electronics and communication electron-
ics are integrated in an overmoulded Small Outline Integrated Circuit (SOIC)
compatible Surface-Mounted Device (SMD) package as shown in Fig. 1 . The
overmould includes the necessary openings towards the opto-electronic devices

and geometries to allow an accurate, rugged and standardized way of mating the
POF ferrules with the package. A special spring mechanism keeps the fiberferrule
positioned vertically and allows for a soft plug-in during assembly and excerts
sufficient holding force to confirm a stable connection of the fiber.
In this contribution we describe our research towards an optical coupling scheme
for the RCLED-POF interface to ensure -3.5 dBm coupled power into the POF un-
der all operating conditions of the device. This includes an investigation into the
behaviour of the RCLED optical output in terms of power and beam profile under
a wide range of bias currents and ambient temperatures, defining different optical
coupling schemes for the application, their nominal specifications, and synthesiz-
ing the key tolerances on all optical parts using an exhaustive Monte Carlo opti-
misation algorithm on the full three-dimensional (3D) description of the complete
RCLED package and detachable POF ferrule.
SOURCE CHARACTERIZATION
Due to the wide temperature range in which the devices have to function, and the
fact that their behaviour immersed in transparant material was unknown, we set
forth to optically characterise the RCLED under a wide range of operating tem-
peratures(Moens et al., 2008; Vervaeke et al., 2009) in order to define an accurate
source model for optical simulations. In order to characterize the beam profile as
emitted in transparant material, we have mounted a 10 mm diameter anti-reflection
coated half glass ball lens, with a refractive index similar to the immersion ma-
terial, centered on the RCLED active region (measurement setup as in Fig. 5 1).
Rays emitted from the active region will undergo minimal refraction at the curved
lens air interface, since they will intersect the lens surface perpendicularly, ef-
fectively allowing a beam characterisation as emitted in the immersion material.
Table 1 summarises the most important source characteristics under different
operating conditions regarding peak wavelength shift, slope efficiency, total output
power as measured with an integrating sphere, and the far field pattern. We found
the spectral behaviour of the RCLEDs at different operating temperatures and bias
currents to be in good correspondence with literature(Sipil et al., 2000) . At 15 mA
we observed a peak wavelength shift of (0.0340.005) nm/C at 115 C . The far
field pattern (full width half maximum) shrinks with 0.45 % at a bias current of
20 mA with respect to 7.5 mA . We refer to Fig. 4 for the temperature influence
on the beam profile of a typical RCLED. An immersion of the active layer of the
devices in a transparant material will alter the extraction efficiency as well as the
beam profile as indicated in(Dumitrescu et al., 2002) . This can also be observed
in our measurement results in Table 1 . We have determined that variations in the
device performance are the main cause for the variability on the characteristics.

Figure 4: Beam profiles in air at 15 mA bias current.
This data is subsequently used to define a source model for optical simulations
POTENTIAL OPTICAL COUPLING CONFIGURATIONS
High coupling efficiency between the RCLEDs and the POFs is necessary in or-
der to meet the high optical budget, needed for automotive applications given the
maximum optical power obtained from RCLEDs. When considering automotive
applications, high reliability and ultra low cost are the prime requirements. These
requirements severely limit the use of optical packaging techniques between light
sources and optical fibre. To our knowledge only a small number of low cost pack-
aging techniques can be considered: molding of clear epoxy or silicone, ultraviolet
embossing, dispensing of a transparant material, die attach of a ball lens on top
of the RCLED or a combination of ball lens die attach and dispensing, mounting
of a mirror and mounting of a biconvex minilens. These optical systems have to
have the potential to increase the coupling efficiency substantially compared to the
RCLED butt coupled with the POF(Kumar et al., 1995; Nakamaru et al., 2002) .
The list of low cost optical assembly techniques and the aforementioned consid-
erations result in five different configurations for the optical coupling system (see
Fig. 5 ).
To determine the maximum coupling efficiency of each of these potential cou-
pling configurations we used ray-tracing software with an accurate RCLED model
defined by extensive characterization (see previous section). This model consists
of a uniformly emitting disk with an appropriate angular ray distribution. We have
incorporated a realistic POF simulation model to simulate the numerical aperture
and Fresnel losses at the entrance facet. The detector of the simulation model is

Table 1: Source characterisation results. Power at -40 C is influenced by ice crystal formation.
Parameter -40 C Room temperature 115 C

Emitting in air
Wavelength ( nm) 659.9; 9.5 FWHM
Peak Wavelength Shift ( nm/mA) (0.0630.070) (0.0540.043) (0.0340.005)
AcademyPublish.org - Vehicle Engineering

Slope Efficiency ( W /A) (0.0510.006) (0.0630.051) (0.0450.013)
Far Field Pattern (FWHM) ( ) at -20 C : 80.5 98.7 78.2
Output Power at 15 mA ( mW ) (0.4810.197) (3.1920.662) (0.3730.121)
Immersed at 15 mA
Far Field Pattern (FWHM) ( ) 61.8
Output Power ( %) +(19.911.6)
81
Figure 5: Nominal design coupling configurations. Between brackets : [coupling
loss ( dB); coupled optical power ( dBm)] (both at 115 C). (1) Clear molded lens
(CML) or Dispensed lens (DL) [-1.61;-1.83]: molded or dispensed lens in clear
epoxy or silicone material; (2) Ball lens (BL) [-1.08;-1.73]: mounted on centering
pedestal and fixed with epoxy; (3) Buried ball lens (BBL) [-1.19;-1.41]: silicone
covered high index of refraction ball lens; (4) Compound parabolic concentrator
(CPC) [-0.86;-1.52]: reflective hollow mirror; (5) Minilens (ML) [-1.08;-1.73]:
biconvex lens system above RCLED. R: radius of curvature; ZC : centre position
of curvature above emitting surface; ZL : adhesive layer thickness ; nB : immersed
ball index of refraction ; L: CPC length ; : CPC exit angle ; 2a: CPC aperture
diameter ; D: ML diameter ; (Zi ,Ri ): ML vertex positions and radii of curvature.
(1) (2) (3)

POF entrance facet
R R nB R
ZL
ZC ZC
ZL ZC RC LED aperture
(4) (5)
(Z2,R2)
L D
a
(Z1,R1)
2a
positioned inside the core at a large distance of the fibre input facet. Dispersion
and attenuation of the POF material were not taken into account.
Fig. 5 gives an overview of the five coupling configurations with the results
for the coupling loss and resulting coupled power in the POF for the worst-case
source, i.e. at 115 C . We conclude that the optimal system, and the correspond-
ing performances, are not severely affected by a different temperature setting and
we are able to keep the optical power coupled into the POF within the -3.5 dBm
specification for the system. Manufacturing and cost considerations narrowed the
choice of configurations down to two for further analysis: the Clear Molded Lens
(CML) and Ball Lens (BL).
TOLERANCE SYNTHESIS
Tolerances define the allowable combined variations in both geometry and po-
sition of elements and their assembly in a system. A parameter sweep analysis,

whereby each individual perturbation is swept within a sensible range while moni-
toring the functionality of the system, already yields an insight into the challenges
to fabricate and assemble the envisioned opto-mechanical system. In a real-life
situation we expect a perturbations to be randomly distributed and occuring si-
multaneously.
The functionality of the system can be quantified using a merit function, usu-
ally a multiple valued vector, based on measurable specifications, e.g. mechanical,
economical, physical or optical, to be met. Constraints defined by the specifica-
tions will constrain the merit function value. The goal is to maximise the number
of fabricated systems complying with the merit function constraint, or in other
words the probability to obtain a system with the desired functionality from a not
error free assembly of components exhibiting dimensional and geometrical varia-
tions.
As a large body of research on this topic has been conducted in mechanical
design, we set off with a brief overview of the tools available in this research
community. A next section is dedicated to the specific tolerance problems already
solved in the optical research space.
Mechanical Tolerancing
In mechanical tolerance terms we usually have to establish a merit function for
a dimensional constraint or a desired force exerted by two parts on each other.
The latter is usually related with a space clearing between objects. A number of
authors have discussed evolutions in the field and to limit the discussion to the
scope of this work, we only present some highlights.
One way to analyse tolerances is to determine an explicit mathematical rela-
tionship between the merit function and the individual dimensional and geometri-
cal variations within the system. Of course this is only possible for non-complex
assemblies with a reduced number of simple components. Once the explicit merit
function is known it is rather straightforward to determine the tolerances on the
individual dimensions and geometries using a mathematical analysis.
A conventional method to define tolerances of more complex systems is to
look at each parameter that will make up the merit function value separately, along
its allowable variation to keep the merit function within its specification. In this
case we revert to a parameter per parameter (PPP) sensitivity analysis. The effect
of simultaneously occurring variations could be estimated using the root-sum-
squares or the worst-case method. Both methods assume that we can describe a
linear relationship between the n p dimensional and geometrical variations pi of
each component in the assembly and the merit function Y with which we describe
the functionality of the system:

np
Y = i pi (2)
i=1
Assuming that every parameter pi is independent (there exists no correlation

between the pi s), we can estimate the mean and variance 2 of the merit func-
tion as (root-sum-squares method):
N
= i i (3)
i=1
N
2 = 2i 2i (4)
i=1
Hereby we assume that Y follows a normal distribution. Even when the in-
dividual parameters pi with mean i and variance 2i differ from the normal dis-
tribution we know that the central limit theorem dictates that when the parameter
count n p is high enough the merit function Y will exhibit a normal distribution.
By setting a lower LL and/or upper UL constraint on the merit function Y we
are easily able to calculate the expected number P of systems complying with the
constraints by integration of the normal distribution of Y between its constraints:
UL (x)2
1
P= e 22 dx (5)
LL 22
Since we know that from Eq. 4 both and relate to the individual parameter
distributions we are able to redefine each i and i to obtain a desired number or
systems complying with LL Y UL . The routine followed here does not require
extensive simulations. We only have to determine the capabilities of our technol-
ogy to define the parameters pi with a normal distribution and the relationship to
describe Y . Whenever Y exhibits an unfavourable distribution we can revert to
Eq. 4 to set new parameter distribution characteristics.
Both the root-sum-squares and the worst-case method induce a considerable
error. The methods yield respectively a too optimistic setting for the dimensional
and geometrical variations and a too severe setting for the parameters of the sys-
tem. Moreover the arbitrary nature of realistic parameter probability density func-
tions (PDF) describing pi , and the possibly nonlinear and/or implicit form of the
merit function is not accounted for in these methods and requires a more detailed
approach to the construction of the merit function. We must mention that other
metrics based on this method for process yield estimation are known in the field
of mechanical tolerancing(Nigam et al., 1995) .
Different approaches to the definition of a merit function based on tolerances
have been examined and will be shortly introduced here: parameter spaces , offset

solids
and feasibility space(Chase et al., 1991; Guilford et al., 1992) .
One of the first mathematical mechanical tolerance models examined in depth,
is the parameter space model for mechanical tolerancing. Vector loops are com-
prehensive example of a parameter space model for a mechanical assembly. They
describe the tolerance kinematics by defining a number of loops of vectors con-
necting individual degrees of freedom in the assembly (see Fig. 6 ). The resulting
set of equations, equalling zero since all loops have to be closed, can be used to
evaluate a free parameter such as a clearance, contact angle between two surfaces
or desired force to be exerted. In simple cases this set of equations is explicit and
can be solved directly to construct the merit function.
Figure 6: Vector loop diagram of a lens mount. The system is supposed to be

rotationsymmetric.
The abstraction of vector loops from the 3D geometry allows for fast itera-
tion but involves a large number of carefully elaborated equations to be solved.
Especially heavily constrained 3D assemblies will yield a large set of implicit
equations for the different vector loops in the model. For more in-depth material
on this subject, we refer to(Glancy et al., 1999; Gao et al., 1995) .
Offset solid modelling is used to check clearances between components in an
assembly by linear expansion of the individual components along its dimensions.
The clearances themselves can stem from a constraint based on a maximum fric-
tion or normal force, or an assembly requirement. By describing the clearances
as a function of the expansion of components it is possible to calculate the neces-
sary tolerances to comply with the constraint on the merit function. This method

only allows for linear expansion or contraction of the solid but can be upgraded to
estimate angular offsets.
The feasibility space description or variational geometry of tolerances defines
an analytical model of the individual components. This fundamental approach
allows for easy design modification and the usage of free variables to analyse
their effect on the assembly, but involves the simultaneous solving of a large
number of nonlinear equations. The method we will use for the Monte Carlo
optomechanical tolerancing of the optical interconnect module is based on this
approach.
The solution route for the tolerancing problem usually involves statistics to
describe the real-world error PDFs. A key issue for a fast-turn-around time during
the design phase of a mechanical construction is fast evaluation and iteration.
Since robust statistical(Varghese et al., 1996; Nigam et al., 1995)
(Cvetko et al., 1998) methods like the Monte Carlo method prohibit fast iteration,
researchers have sought faster solution methods into the problem. These involve
usually a linearisation of the problem(Glancy et al., 1999) ,
a Design of Experiments approach and genetic algorithms
(Singh et al., 2003).
Meanwhile, tolerancing tools have found their way into Computer Aided En-
gineering software for first validation of dimensional and geometrical parts and
assemblies(Adc, 2011; 3ds, 2011). The common response functions are based on
dimensional and geometrical constraints, load analysis or even a manufacturing
cost estimation of parts and assem-
blies(Glancy et al., 1999).
To our knowledge, no integrated approach has been researched using mechan-
ical tolerancing using other merit functions, e.g. Modulation Transfer Function
(MTF), link efficiency or aberration figures from optical simulations. We there-
fore decided to implement a framework for the opto-mechanical design and anal-
ysis of tolerances.
Optical Tolerancing
Tolerancing efforts in optics, and more specifically optical interconnects, is usu-
ally restricted to a sensitivity analysis approach of the designed system. Although
defining a workable tolerance range is imperative for the fabrication and assem-
bly of optics in real-world systems, only few authors addressed the implications
of multiple misalignments on optical performance characteristics of interconnect
modules. We emphasize here on work done in the optical interconnect commu-
nity because of the required accuracies and miniaturisation of interconnects base
at this interconnect hierarchy level.

One study uses a n 4 f processor architecture to evaluate the tolerances of a
free-space interconnect(Kirk et al., 2003) . The interconnect model used here is
an n 4 f multistage relay system. 2 rotational errors and 9 translation errors on
each 4 f system are swept in a random manner according to normal distributions.
The authors conclude that the realistic combination of assembly errors greatly
affect the probability to fabricate passively assembled systems meeting specifica-
tions. Moreover a sensitivity analysis is not adequate to set the tolerance limits
because of the lack interaction of errors. The authors also conclude that nor a root-
sum-squares nor a worst-case method estimates the process yield accurately. The
investigation of this optical interconnect system also reveals a significant decrease
in process yield with increasing system complexity.
Other authors used regression analysis to define the merit function
(Ozkan et al., 2002) . By analysing the contribution of different parameters to the
merit function model it is possible to eliminate non-significant parameters and to
assess correlations between parameters. The merit function can be used to predict
the process yield and is suitable for fast synthesis of the tolerances, but relies on
simplifications regarding the stochastic nature of parameter perturbations.
Also, full-scale link simulators(Levitan et al., 1998) and a smart-pixel packag-
ing scheme(Morozov et al., 1996) have been reported in literature but in general,
none of these efforts have dealt with the specific tolerance problems coming with
a realistic three-dimensional packaging scheme for free-space intra-chip optical
interconnects.
Although some optical simulation tools already provide an interface for toler-
ancing, or have been extended to encompass Monte Carlo based tools(Karppinen
et al., 2004) , this is usually limited to the core of the optical structure, and not
the underlying mechanical alignment features. The main reason for the lack of in-
tegrated simulation engines is the complexity level and the absence of integrated
opto-mechanical design and analysis tools enabling a full tolerance analysis.
Monte Carlo Simulation Principles

The Monte Carlo method is a statistical sampling technique that has been applied
in numerous fields. It originates from an idea of Stan Ulam when playing solitaire.
He wondered whether mere observation of a large number of experiments would
yield more practical method to obtain value for the number of successful plays
of solitaire than using combinatorial calculations. (Eckhardt, 1987). John Von
Neumann took up this idea and explored statistical sampling in a totally different
domain than the already known mathematical context: the estimation of neutron
multiplication rates.
Both the complexity of an analytical representation of a system-under-test and

the inherent stochastic nature of one or more parameters defend the application of
Monte Carlo methods. The random nature of errors and their stack-up in a man-
ufacturing process sets the rationale for using Monte Carlo simulation to evaluate
their implications on fabrication accuracy, assembly ease and ultimately produc-
tion cost.
The ability to generate random numbers following an arbitrary probability
density function (PDF) is a key element of the Monte Carlo method. Several
methods to generate a population following a predefined PDF have been docu-
mented (Fishman, 1996). These random numbers are subsequently attached to
parameters of the system-under-test. Simulation of the system yields a set of per-
formance values. When enough experiments are carried out, the experimenter can
calculate statistical characteristics of the obtained sets of performance values.
We laid out in a previous section that the combination of a mechanical and
optical computer aided design tool would improve the whole design process of
optical interconnect systems. Instead of creating in a stepwise manner the defi-
nition of the final assembly using subsequent optical and mechanical design, we
would be able to approach the problem in a holistic fashion. The absence of a
true integrated tool for opto-mechanical design sets our goal in this and following
sections.
To perform the statistical analysis of the fabrication and assembly process, we
have to define a response function to analyse the performance of the system. We
decided to use the process yield, defined as the number of systems complying
with a given set of limits on transmission loss and cross-talk, as the response
function from the Monte Carlo method. The process yield can also be viewed
in terms of a probability to fabricate and assemble systems within specification.
The larger the value of the process yield, the lower the fall-out and the higher the
total production efficiency will be, which in its turn has an effect on the cost-per-
component. Of course, one has to be careful to consider an appropriate set of
tolerances to balance the fabrication cost.
We applied the Monte Carlo method on the optical interconnect module. A
large batch of geometries of the module are defined, each with a slightly different
set of misalignment errors. The Monte Carlo engine drives an optical simulation
tool in which we model the interconnect module. The optical simulator returns the
optical characteristics of every system in the Monte Carlo batch. By performing
a statistical analysis on the obtained optical characteristics, we are able to assess
the assembly and fabrication yield and set other tolerances for the individual mis-
alignments.

Mechanical Model of the MOST Transceiver Module
We have implemented the geometry of the RCLED, its package, the detachable
ferrule and the POF sitting inside it, as well as the optical system as described
above in a realistic mechanical 3D simulation. This simulation environment is
able to model dimensional and geometrical perturbations of the complete assem-
bly to assess the resulting optical performance. The model is based on a polygon
description of the mechanical and optical structures along with algorithms to de-
tect and resolve collisions and is implemented in Matlab and Zemax(Vervaeke
et al., 2006b) .
The polygon description of the optical interconnect module we implemented
consists of three vectors for each separate face of a component. One vector de-
scribes each cornerpoint, or node, of the part while a second a vector describes the
connections, also called vertexes, between each point. The last vector we used de-
fines each face by assigning a collection of vertexes belonging to that face, which
we assume to be flat and not subject of any deformation causing a local curvature.
As for spherical or rotational symmetric shapes, we opted to provide an alternative
description to avoid errors due to polygon sampling of the shape. We describe a
sphere by its centrepoint coordinates in the node vector, and a vector setting its ra-
dius. It is clear that with this methodology to describe a three-dimensional object
we can easily design operators to expand or contract a part, apply rotations and
translations or arbitrary deformations as long as no local curvature occurs on an
isolated face.
To be useful in a Monte Carlo simulation scheme, we have implemented a
number of tools to detect and resolve collisions between components in an as-
sembly. By applying arbitrary geometrical and dimensional perturbations on the
system, it might occur that faces of parts are intruding, which is not reflecting re-
ality where a face might stop the free movement of a part. Moreover: we want to
simulate the assembly process as accurately as possible. This means that we need
to detect when a part is mating its counterpart and how it is positioned relatively
to another component.
The procedure followed uses the full vectorial representation of the objects(Jimenez
et al., 2001(@) (see Fig. 7 ) whereby each polygon face is evaluated for intrusions
into another polygon face using elementary vector mathematics. Each face Fqr of
the object Oq is represented by its determinant description:
Dqr =
pqrs
n
qr (6)
wherein pqrs
represents a cornerpoint belonging to the face and n
qr the face
normal of face Fqr . A face intrusion of a polygon vertex vi jk = pi jk+1

p
i jk be-
longing to face Fi j of object Oi with face Fqr of Oq is detected by calculating the

Figure 7: Full vectorial intrusion detection layout.

intrusion vector Ii jkqr along the polygon vertex:

Ii jkqr = pi jk + Ri jkqr v
i jk (7)
whereby the location along vertex
v
i jk of the intersection point is determined
by Ri jkqr :
Dqr
n
qr pi jk
Ri jkqr = (8)
nqr vi jk
The latter equation stems from the fact that the intrusion vector must belong
to the face Fqr , which can be expressed as:

Ii jkqr Fqr
n
qr Ii jkqr = Dqr (9)
The last part of the collision detection algorithm we implemented checks
whether an intrusion vector is lying inside or outside the region defined by the
vertices of the face Fqr it crosses by calculating the direction of following cross-
product at each face point pqrs
:

XIs = (
pqrs Ii jkqrs ) (
p
qrs+1 Ii jkqr ) (10)
If all vectors XIs are pointing the same way, we may conclude that Ii jkqr is
lying inside the face Fqr of object Oq . When the additional condition that the
fraction Ri jkqr 1, we accept the intersection vector Ii jkqr as an intersection of
vertex vi jk Fi j of object Oi with face Fqr of Oq .
To remove the intrusion of face Fi j , it is sufficient to calculate the projection of
the translation vector along the normal of the face Fqr and select the appropriate
direction to resolve the conflict. Finally, we calculate the displacement vector
by selecting the largest vector for each face Fi j intrusion into Fqr and an average
vector to remove the collision of object Oi with Oq . The displacement itself is
limited by a predefined factor and if needed, the collision algorithm is rerun until
the residual collision is lower than a preset limit (currently 1 nm).

Tolerancing of the MOST Transceiver Module
We use a Monte Carlo optimisation scheme to determine the optimal tolerances on
the optical coupling elements to ensure all specifications Si , in our case minimum
-3.5 dBm coupled power in the POF, are met with a sufficient probability. We
therefore will optimise the process yield as a function of a given parameter set p:
P = f (p). By generating N systems with random perturbations we are able to cal-
culate the process yield using a finite sum: P = f (p) = 1/N Nj=0 (p j ) whereby
(p j ) = 1 if for a parameter set p j all specifications Si are met, and (p j ) = 0 oth-
erwise. We define three classes of parameters sets: deterministic parameters p[d]
exclusively defining the optical system and hence directly related to the optimal
optical coupling efficiency, stochastic parameters p[s] describing all other pertur-
bations on dimensions and geometries of the elements and the assembly and p[]
defining the operating conditions. The optimisation of P for a certain target value
P is carried out by selecting each iteration j the parameter sets p[d] j describing
the optical coupling scheme leading to a system compliant with all Si . The mean
and variance of the corresponding Gaussian Probability Density Functions (PDF)
d ([d] , 2[d] ) are recalculated from the selected parameter set using a running av-
erage and running variance scheme. We stop the algorithm when the likelihood to
have a non-compliant system (i.e. < -3.5 dBm coupled power) drops to 106
(see Alg. 1). Next to the deterministic parameters p[d] of the optical coupling
scheme (in our case R, ZL and ZC for the BL as in Fig. 5 ), we have two other cat-
egories of parameters: static parameters p[s] with fixed PDFs [s] describing the
geometries and dimensions of the assembly, and operating conditions p[] . p[] is
the worst-case beam total power and beam divergence, i.e. at 115 C .
The algorithms described above were subject of earlier research
(Vervaeke et al., 2006b,a; Vervaeke, 2007) on optical interconnects for on-chip
signalling. Fig. 8 shows a multiparameter Monte-Carlo tolerance synthesis it-
eration step wherein 11 parameters of an optical interconnect are subject of an
in-depth analysis. The geometrical and dimensional variations on the complete
package stack of this interconnect are based on manufacturing and assembly data
and lead to the histograms shown. The number of systems capable of reaching a
determined transmission efficiency the process yield in this example for the op-
tical interconnect is depicted in Fig. 9 . Cutting off all systems with an undesired
low transmission efficiency leads to the white histograms of Fig. 8 . Hence we are
able to recalculate the distributions on the 11 free parameters in the system to start
another simulation round. The goal is to increase the process yield of Fig. 9 to a
desired value. The algorithm for the MOST transceiver is very similar except that
each individual system generated is evaluated and that the total number of free
parameters is limited to three.

Figure 8: Multiparameter Monte-Carlo tolerance synthesis: 11 initial (black) ver-
sus new (white) parameter histograms after one iteration. ni indicates the width
of the selected histogram with respect to the original (ni = 3). The value of <>
indicates the difference between the original mean (set to zero) and the new mean.
As can be seen it is desirable to tighten the specifications for the linear parameters
and shift the nominal design point of working distance and focal length.

Algorithm 1 Algorithm 1: Monte Carlo optimisation with running PDF scheme
1: Set p[d]1 from the nominal design
2: Set d using a parameter sweep and/or from other technological data
3: Set s using a PPP sweep
4: P0
5: j0
6: while P P do
7: Generate a random parameter set p j according to pd , d , s and
8: Calculate (p j )
9: if (p j ) = 1 then
10: Add parameter set p j to the dataset of valid parameter samples
11: Calculate pd , d , v using a running average and variance scheme
12: else
13: Reject parameter set p j as a valid sample
14: end if
15: j = j+1
16: Calculate P = f (p) = 1/N Nj=0 (p j )
17: end while
18: Full statistical analysis of the end result
Figure 9: The process yield evolution versus the set transmission loss limit for a
-3 dB design at i = i /3. 50 % process yield can be expected when setting the
transmission efficiency to 53 % or -2.7 dB.

Figure 10: Melexis MOST 150 SMD MLX75605 package body (blue line) and
ferrule model (red).

Table 2: (3 ) tolerances for two configurations.
Parameter Ball Lens Clear Molded Lens

ZC ( m) 145 4.4 179 14.4
R ( m) 138 5.1 167 8.8
ZL ( m) NA 174 18.9
Applying the Monte-Carlo based tolerance synthesis method on the 3D geo-

metrical model (see Fig. 10 ) of the MOST 150 SMD we were able to synthesise
the tolerances on the optical coupling system (see Table 2 ). The initial values
for all deterministic parameters p[d]1 were set to [d] (1 [d] ) =(150 50) m
. Acceptable results are reached after 105 iterations. The CML and DL configu-
rations are less critical regarding the radius of curvature R, centre position ZC and
adhesive layer thickness ZL . For these techniques the position of the lens vertex
with respect to the conical seat on the package, intended to fit the POF ferrule,
is fixed (CML) or has a small tolerance (DL). The average coupled power for the
worst-case source is -2.7 dBm for both systems. The BL configuration was cho-
sen considering the total cost for manufacturing and assembly with respect to the
tolerances required to meet the power specification.
CONCLUSION
The automotive industry is taking up the success of POF to the benefit of high-end
datacommunication systems in automobiles. The competitive market and strong
specifications pose challenges regarding robustness and alignment tolerances to
any optical design incorporated in the transceivers for the MOST network. The
tolerance synthesis approach for complex opto-mechanical systems described here
enabled us to select two out of five candidate optical coupling systems, and to fully
simulate their behaviour under realistic 3D dimensional and geometrical perturba-
tions that occur during fabrication and assembly of the various components of the
transceiver and the POF ferrule. The study enabled us to determine the parame-
ters of the optical system to ensure a maximum coupling efficiency at all operating
conditions using a source model derived from measurements on the RCLED beam
characteristics. Including data from production and cost models we have been able
to identify the ball lens configuration scheme as the most cost effective in view of
the tolerances needed for manufacturing and assembly.

ACKNOWLEDGEMENTS
This work was supported in part by DWTC-IAP, FWO, GOA, the 6th FP European
Network of Excellence on Micro-Optics NEMO, the Methusalem and Hercules
foundations and the OZR of the Vrije Universiteit Brussel (VUB).
References
http://www.3ds.com/products/catia/weclome/. Website Dassault Systems. URL

http://www.3ds.com.
http://adcats.et.byu.edu/home.php. Website Association for the De-

velopment of Computer Aided Tolerancing Systems. URL
http://adcats.et.byu.edu/home.php.
K. W. Chase and A. R. Parkinson. A Survey of Research in the Ap-

plication of Tolerance Analysis to the Design of Mechanical Assemblies.
Research in Engineering Design, 3:2337, 1991.
R. Cvetko, K. W. Chase, and S. P. Magleby. New metrics for evaluating Monte

Carlo tolerance analysis of assemblies. http://adcats.et.byu.edu.
W Daum, J Krauser, PE Zamzov, and O Ziemann.

Polymers Optical Fibers for Data Communication. Springer, 2002.
MM Dumitrescu, MJ Saarinen, MD Guina, and MV Pessa.

High-speed resonant cavity light-emitting diodes at 650 nm.
IEEE Journal Of Selected Topics In Quantum Electronics, 8(2):219230,
Mar-Apr 2002. ISSN 1077-260X.
R. Eckhardt. Stan Ulam, John von Neumann, and the Monte Carlo method.
Los Alamos Science, (15):131137, 1987.
G. S. Fishman. Monte Carlo: Concepts, Algorithms and Applications. Springer

Verlag, 1996.
J. Gao, K. W. Chase, and S. P. Magleby. Comparison of assembly tolerance analy-

sis by the Direct Linearization and Modified Monte Carlo Simulation Methods.
In Proc. of ASME Design Engineering Technical Conference, pages 353360,
September 1995.
C. G. Glancy and K. W. Chase. A second-order method for assembly tolerance

analysis. ASME, September 1999.

Andreas Grzemba, editor. MOST: The Automotive Multimedia Network. Franzis
Verlag GmbH, Poing (D), 2011.
J. Guilford and J. Turner. Representing geometrical tolerances in solid models.

Computers in Engineering, 1:319327, 1992.
P. Jimenez, F. Thomas, and C. Torras. 3D Collision Detection: A Survey.

Computers and Graphics, 25(2):269285, April 2001.
M. Karppinen, J. Makinen, K. Kataja, A. Tanskanen, T. Alajoki, P. Karioja, M. Im-

monen, and J. Kivilahti. Embedded optical interconnect on printed wiring
board. volume 5453, pages 150164, 2004.
A. Kirk, D. V. Plant, M. H. Ayliffe, M. Chateauneuf, and F. Lacroix.

Design rules for highly parallel free-space optical interconnects.
IEEE Journal of Selected Topics in Quantum Electronics, 9(2):531547,
MarchApril 2003.
A Kumar, A Suzuki, K Kurihara, F Miyasaka, H Hotta, and K Sugita. High-

brightness, AlGaInP-based, visible light-emitting diode for efficient coupling
with POF. IEEE Photonics Technology Letters, 7(10):11341136, Oct 1995.
ISSN 1041-1135.
S. P. Levitan, T. P. Kurzweg, P. J. Marchand, M. A. Rempel, D. M. Chiarulli, J. A.

Martinez, J. M. Bridgen, C. Fan, and F. B. McCormick. Chatoyant: a computer-
aided-design tool for free-space optoelectronic systems. Applied Optics, 1998.
E. Moens, M. Vervaeke, Y. Meuret, H. Ottevaere, C. Van Buggenhout, and

P. De Pauw. Realistic opto-mechanical modelling of plastic optical fiber. In
SPIE Proceedings Conference on Micro-optics, volume 6992, April 2008.
Valentin N. Morozov, Yung-Cheng Lee, John A. Neff, D. OBrien, Timothy S.

McLaren, and Haijun Zhou. Tolerance analysis for three-dimensional opto-
electronic systems packaging. Optical Engineering, 35(7):20342044, 1996.
F Nakamaru, Y Matsumoto, and A Nakazono. Novel high-efficiency concentrator

for optical fiber communication. IEEE Photonics Technology Letters, 14(7):
953955, Jul 2002. ISSN 1041-1135.
S. D. Nigam and J. U. Turner. Review of statistical approaches to tolerance anal-

ysis. Computer-Aided Design, 27(1):615, 1995.
N. S. F. Ozkan, W. L. Hendrick, P. J. Marchand, and S. C. Esener. Misalignment

tolerance analysis of free-space optical interconnects via statistical methods.
Applied Optics, 41(14):26862694, May 2002.

P. K. Singh, P. K. Jain, and S. C. Jain. Simultaneous optimal selection of design
and manufacturing tolerances with different stack-up conditions using genetic
algorithms. International Journal of Production Research, 41(11):24112429,
2003.
P Sipil, M Saarinen, M Guina, V Vilokkinen, M Toivoren, and M Pessa. Temper-

ature behavior of resonant cavity light-emitting diodes at 650 nn. 15:418421,
2000.
P. Varghese, R. N. Barswell, B. Wang, and C. Zhang. Statistical tolerance analy-

sis using FRPDF and numerical convolution. Computer-Aided Design, 28(9),
1996.
M. Vervaeke. Photonic modelling and packaging of chip-level interconnects. Phd

thesis, Vrije Universiteit Brussel, June 2007.
M. Vervaeke, M. Lahti, M. Karpinnen, C. Debaes, B. Volckaerts, P. Kar-

ioja, and H. Thienpont. Packaging a free-space intra-chip optical inter-
connect module: Monte Carlo tolerance study and assembly results. In
Micro-Optics, VCSELs, and Photonic Interconnects II, volume 6185 of SPIE,
page P1850. 2006.
M. Vervaeke, C. Debaes, B. Volckaerts, and H. Thienpont. Optomechanical Monte

Carlo tolerancing study of a packaged free-space intra-MCM optical intercon-
nect system. IEEE Journal Of Selected Topics In Quantum Electronics, 12(5):
988996, Sep-Oct 2006.
M. Vervaeke, E. Moens, Y. Meuret, H. Ottevaere, C. Van Buggenhout, P. De Pauw,

and H. Thienpont. Tolerance Design of an Optomechanical Transmitter Assem-
bly for Automotive Applications. IEEE Photonics Technology Letters, 21(17):
11781180, 2009.

Modeling Window Glass Automotive Antennas with EM
Solvers. Experimental Validation Using a Full-Sized Car
Model
Dr.Matteo Cerretelli, Dr.Paolo Facchini, Dr. Andrea Notari ,
Prof. Guido Biffi Gentili
INTRODUCTION
Automotive antennas need to satisfy various stringent requirements. Since vehicle mount
antennas are mostly installed in confined spaces, small size and lightweight are natural
requirements; then low-profile, flush-mounting and conformability of the antenna
elements are often required in modern automobiles (Nishikawa, 2003; Cerretelli and Biffi
Gentili, 2007).
Automotive antennas are never used in isolation, because they are embedded in the
complex car body environment with its associated internal and external items. Therefore,
even if the isolated antenna would have an omnidirectional radiation pattern, fluctuations
of the received voltage are expected due to the electromagnetic (EM) coupling between
the antenna and its environment.
Window glass antennas are normally composed by narrow strips printed on the internal
face of the glass using highly conducting ink. The EM coupling with the vehicle window
aperture is particularly strong in this case and then radiation currents will be induced on
the surface of the car shell that will behave as a distributed secondary radiator.
Particularly, in the FM frequency band, the size of the window aperture is of the order of
half a wavelength. Consequently strong and non-uniform currents can localize on its
edges.
Furthermore, because the FM free-space wavelength is about 3 meters, the car body is
near the resonant region, therefore small frequency variations could cause considerable
current and radiation patterns distortions with respect to the isolated antenna case. For
these reasons window glass antennas have to be designed taking into account the
influence of the vehicle shape and the presence of internal metallic structures such as
front and back seats.
Numerical analysis is very useful for this purpose, and CAD (Computer Aided Design)
by computer simulation was widely used in recent years in conjunction with experiments.
It is worth noting that measurements in a far field range are highly burdensome and time
consuming, then they could be confined only to the final testing phase, owing also to the
fact that the complete design cycle of a new car is nowadays more and more shortened in
time. Furthermore, because the design of the new antenna is not made by the car producer

but by a specialized antenna firms before the actual car is available, very reliable and
accurate EM simulators become essential for the design optimization phase.
In principle the EM modeling of an antenna printed or embedded on the window glass of
a car can be faced-up using both frequency domain and time domain full-wave solvers.
The advantage of the frequency domain MOM (Method of Moments) solver (Rao, Wilton
and Glisson, 1982) is that it is a source method, where only the car shell is discretized.
Problems could arise in discretizing non-metallic parts as glass windows, sunroof, plastic
bonnet and moldings. Nevertheless the effect of dielectric layers can be accounted for by
introducing equivalent electric and magnetic currents on the surface of the plastic bodies.
Against the best point of generality, the disadvantage of the time domain solvers as
FDTD (Finite Difference Time Domain) (Taflove and Brodwin, 1975) or FIT (Finite
Integration Technique) (Weiland, 1977) methods are intrinsic in their nature of being
field methods, meaning that the complete volume domain of the problem, including
free space, must be discretized and ABC (Absorbing Boundary Conditions) (Mur, 1981)
such as PML (Perfectly Matched Layer) (Berenger, 1994) must be introduced to study
open space radiation.
At present both solving methods have been fully developed and accurately refined in
order to treat more and more complex problems, particularly in the aerospace and defense
fields, such as RCS (Radar Cross Section) optimization and performance analysis of
communication/radar antennas on airplane or satellite bodies.
Despite the availability of these very powerful numerical solvers, their accuracy in
simulating and optimizing glass antennas on a real car remains questionable, in fact up to
now the published results (Abou-Jaoude and Walton, 1998; Low, Langley and Batchleor,
2007; Low, Langley, Breden and Callaghan, 2006) often show a scarce adherence between
numerical and experimental results.
The causes of this lack of consistency are related to a multiplicity of factors as the high
complexity of the car body and its internal parts, the difficulty to model the environments
of the antenna feed and finally the presence of the driver and the passengers.
A typical problem that arises in the full-wave EM analysis of embedded automotive
antennas is the poor mechanical geometry and fault intolerant meshing when importing
drawings coming from the car firms. Usually the original CAD models (typically 50-200
MB in CATIA or Unigraphics format) cannot be directly employed for the EM numerical
analysis because they include too much fine details to be discretized causing the
explosion of the simulation run time (especially for the MOM approach). In addition
importing and managing the model also take a long time. A bad consequence of the fault
intolerant meshing is the hidden removal of the existing air gaps between different parts
of the bodywork (around the doors, for example) or the insertion of false breaks that
locally produces erroneous superficial current behaviors and poor estimation of the
antenna radiation patterns.
The procedure of reducing the complexity of the original car model and locally remediate
corrupted geometrical descriptions is difficult enough to be completely automated. At
present it is performed in some steps in order to reduce the file dimension of at least one
order of magnitude and to reach a coherent model that can be handled by the EM
simulator.

It is worth observing that the prediction of the angular position of the nulls in the
radiation patterns is very difficult because nulls are extremely sensitive to the current
distribution on the car shell and on the internal furnishings.
Therefore the accuracy in predicting the radiation patterns nulls on both horizontal and
vertical field components could be assumed as an index of the truthfulness of the EM
simulation. In this paper, we present a study of the accuracy that can be achieved in
modeling FM window glass antenna by employing two of the more powerful and popular
3D EM field simulators commercially available: FEKO, by Electromagnetic Software &
System-SA, and CST Microwave Studio, by Computer Simulation Technology GmbH.
The former, being based on the source method approach, is a MOM integral equation
solver in frequency domain, while the latter, based on the field method concept, is a
FIT time domain solver, associated to the proprietary Perfect Boundary Approximation,
PBA (Krietenstein, Schuhmann, Thoma and Weiland, 1998).
The MOM solver has been universally recognized, in principle, as the more efficient
method to analyze very large metallic structure, particularly by resorting to the alternative
Multilevel Fast Multipole Method (MLFMM) (Engheta, Murphy, Rokhlin and Vassiliou,
1992; Song, Lu and Chew, 1997). Nevertheless some recent advances in the FIT solver, as
flexible and adaptive subgridding, have allowed increasing the numerical efficiency of
this method in analyzing complex and large objects. Because a car has dimensions of few
wavelengths in the FM band, both methods are good candidates for the EM full-wave
simulation of FM antennas in a vehicle.
COMPARISON BETWEEN MEASURED AND SIMULATED RESULTS
The overall accuracy obtainable in the numerical modeling of the embedded automotive
antennas using the available EM solvers can be estimated by comparing numerical results
to the measurements performed in a suitable open site.
For this purpose it is not convenient to utilize a real car as reference platform because we
are not able to a priori quantify the EM modeling uncertainties introduced in reducing the
complexity of the original structure to create a manageable EM model.
Fig.1. Mechanical drawing of the phantom car model (units millimeters; car width=1500,
not shown).

It is therefore preferable to test the modeling accuracy using a stylized car model as that
depicted in (Fig. 1). This model has dimensions comparable to a real commercial car, but
is built with a stylized shape obtained by a combination of planar surfaces, having
removed the small details that are not significant from the electromagnetic point of view.
This simpler shape was also chosen in order to be easily implemented in any numerical
code without further structural simplifications, thus avoiding the uncertainties that come
with the description of a real car (dielectric materials not well characterized, external
details removed to reduce numerical effort, inconsistency between CAD data and real car
and so on). The physical model of the stylized car, shown in (Fig. 2), has been fabricated
using a wood framework covered with a conductive tissue, characterized by a resistance
of 0.04 Ohm per square. The roof and the top of the trunk are instead realized by a
galvanized 2 mm thick plate to assure a satisfactory mechanical fix of the standard
monopole used as reference antenna.
Fig. 2. Photo of the realized car model upon the rotating platform.
The antenna under test, a simple copper wire monopole with a diameter of 0.4mm, was
fixed to the internal glass surface of the littlest side window as depicted in (Fig. 3).

Fig. 3. Layout of the antenna under test. The feeding point is 50mm below the bottom
edge of the window.
In the measurement site the stylized physical model of the car is placed on a rotating
platform. A far transmitting log-periodic antenna generates the wavefront that impinges
the car model supporting both the window antenna under test and a standard reference
monopole. By properly setting the transmitting antenna height and distance from the
rotating platform the radiation patterns is measured at an elevation of 5.8. The metallic
rotating platform (diameter 4.5m) and the surrounding real ground having r=3 and
=0.0001 S/m, corresponding to dry ground as in von Hippel (1954), are included in the
simulations in order to correctly take into account the environmental effects.
In FEKO this real ground was modeled by using the Green function of a stratified media,
while in CST Microwave Studio it was considered as a stratification of the background
material, thus applying the decoupling plane feature of the farfield visualization. This
implies that the comparison between the computed patterns of FEKO (gain patterns) and
CST (farfield patterns) can be done only by referring to the relative field levels.
Using the car model of (Fig. 1) with the antenna of (Fig. 3) the radiation patterns of the
embedded antenna have been computed with both FEKO and CST solvers.
In CST the glass of the window was modeled as a solid having the same shape, thickness
(4mm) and dielectric properties (r=7, tan=0.02) of the real glass.
In FEKO the glass antenna loading was modeled by adding a coating to the wire using
the card CO in EditFEKO. This approach uses an analytical correction of the currents on
the wire and has the advantage of not affecting the simulation time. The thickness of the
equivalent coating was chosen to be double (8mm) of the glass thickness while for r and
tan the real values was used, according to calibration experiences and to a FEKO
application note (Jakobus, Berger and Landstorfer, 2000).

Fig. 4. Measured radiation pattern (continuous line) at 86MHz, 96MHz and 106MHz
(dB) for vertical (up) and horizontal (down) polarization compared with FEKO (dotted
line) and CST (dashed line) simulations. Elevation angle 5.8.
86MHz 96MHz 106MHz
Vertical Polarization
86MHz 96MHz 106MHz

Horizontal Polarization
(Fig. 4) shows the comparison between the measured and simulated patterns at the
starting, center and ending frequency of the European FM band, for both vertical and
horizontal polarization for a horizontal cut corresponding to the 5.8 elevation angle
achievable in the measuring open area test site. Since the measured levels, retrieved in
dBV, are dependent on the power transmitted by the log periodic antenna, the distance
between transmitting and receiving antennas and the attenuation in the connecting cables,
all the curves are given normalized to the measured patterns, in an absolute dB scale for a
better comparison.
Both CST and FEKO simulation results agree well with the measured data also at the
extremes of the FM band. CST Microwave Studio required 500Mb RAM and 1.75 hours
on a Intel Core 2 Duo E6600 processor for a wideband result (50 200MHz), while
FEKO required 260Mb RAM and 2.35 hours for calculating 8 frequency points.
It is worth nothing that the nulls of the radiation pattern are well predicted with accuracy
better of +/- 10 from both numerical solvers thus confirming that a strict consistency
between the EM model and the original CAD model is essential for obtaining reliable
simulation results.
(Fig. 5) shows a comparison of the input reflection coefficient for the measured antenna
and the simulations from the two EM solvers used. Both numerical solvers show some
differences from the real model; the MoM solution with FEKO shows a differences in the
resonant frequency while the amplitude is well predicted whereas the FIT simulation with
CST Microwave Studio show greater agreement in the main resonant frequency but at the
same time show a second resonant that is not experimented on the real model. These

differences may be explained considering that both the solvers have some problems when
simulating the input impedance of a window glass antenna because the difficulty of
accurately modeling the junction between the radiating element and the feeding coaxial
cable. A possible solution to this problem could be the separate simulation of the junction
and its subsequent re-insertion as a black-box device in the final simulation run.
Fig. 5. Input reflection coefficient of the antenna. Measured, simulated with FEKO and
simulated with CST.
INFLUENCE OF STRUCTURAL DETAILS AND INTERNAL FIXTURES
The stylized car model of (Fig. 1) is very useful for performing a numerical
perturbational analysis of the original structure. We can add in sequence some external or
internal structural details or fixtures and analyze their influence on the antenna
performances. For instance if we disconnect the door borders from the car body and leave
only two connection points (hinges), the surface currents are locally perturbed and pattern
nulls can be influenced.
In this section we use the CST simulator to study how the patterns of the antenna under
test are affected by the internal fixtures as the front seats. For better realism also the
presence of the driver is taken into account.
In (Fig. 6) the RF surface current distribution at 96MHz center frequency is shown for
three different situations: the empty car, the car with seats included and the car including
both seats and driver. The seats have been modeled with a brick directly connected to the
vehicle bottom (W x L x P 0.50m x 0.55m x 0.20m) and superimposed by a rectangular
face 0.50m wide and 0.52m high, representing the back of the seat. The driver has been
modeled as a homogeneous stylized phantom filled with an equivalent biological tissue
with r=47 and tan=0.6553 (muscle tissue).

Fig. 6. The car model layout comprising the front seats and the driver (a). Simulated
surface current distribution at 96MHz for: the empty car (b); the car with only front seats
(c); the car with front seats and driver (d).
a) b)
c) d)
As discussed before in CST Microwave Studio, when taking into account the effects on
patterns of an infinite extension of the real ground plane, the use of the decoupling
plane feature is compelled. Since this feature does not allow to compute the absolute
antenna gain, but only the nearfield E field pattern, in this section graphs have been
represented using an absolute dB levels scale, after removing the dependence on the
transmitted power level and on the distance of observation. Therefore, in (Fig. 7) the
radiation patterns produced by the aforementioned three different car configuration are
compared, in an absolute dB scale, for both vertical and horizontal polarization at
86MHz, 96MHz and 106MHz.

Fig. 7. Simulated radiation pattern at 86MHz, 96MHz and 106MHz (dB) for vertical (up)
and horizontal (down) polarization at an elevation angle of 5.8. Empty car (continuous
line), with seats included (dashed line) and with seats and driver included (dotted line).
86MHz 96MHz 106MHz
Vertical Polarization
86MHz 96MHz 106MHz

Horizontal Polarization
At 86MHz the seats show a resonance that deeply affects the pattern but at the same time
fades when considering the presence of the driver, whose dielectric losses reduce the
resonances effects making the pattern look like in the empty car case. At 96MHz and
106MHz instead the behavior is quite similar, with the presence of the seats that affects,
as expected, the radiation patterns of the antenna, while the presence of the driver is here
quite not noticeable.
CONCLUSIONS
In this article we dealt with the matter of the accuracy achievable by employing two of
the more widespread commercial EM full wave solvers for the simulation of window
glass automotive antennas.
A reduced complexity full scale model of a car with a wire antenna fixed to the glass
surface of a side window has been constructed for the measurements in an open site. In
parallel EM simulations have been performed using both solvers.
From the examination of the radiation patterns behaviors we can assert that both
simulators, although based on very different solver engines, provide results that are in
good agreement with the measurements. The angular positions of the radiation patterns
nulls or minima in both vertical and horizontal polarization are independently well
predicted by each simulator with an error of about +/- 10 that represents the best result
up to now reported in the literature. The depths of the minima do not agree well because
the measurement is affected by multipath phenomena, very sensitive to the environmental
conditions in an open space measuring site. After the validation of the accuracy of the

numerical solvers the influence of internal fixtures on the antenna radiation patterns has
been analyzed.
Results of the perturbational analysis show that the surface current distribution on the car
body is very sensitive to the presence and position of internal and external details and
fixtures, as well absorbing bodies. This high sensitivity represents a serious limit of the
methodology because in most practical cases the consistency of the model to the real car
structure cannot be guaranteed.
Some difficulties in performing time and memory efficient numerical simulations are
expected with more involved glass antennas and feeding systems in the real environment.
Future work will concern the analysis of the influence of the fault intolerant meshing of
the real vehicle, consisting in the hidden removal of the existing air gaps between
different parts of the car bodywork (around the doors, for example) or the insertion of
false breaks that locally produces erroneous superficial current behaviors that after all
might influence the antenna radiation patterns.
On the whole it emerges the feeling that the EM simulation of a FM window antenna, if
performed with the necessary shrewdness, is substantially reliable and can constitute a
powerful tool for the design of a window glass antenna.
REFERENCES
Abou-Jaoude and Walton, USA (1998). Numerical Modeling of On-glass Conformal

Automobile Antennas, IEEE Trans. Antennas and Propagation, Vol. 46, No.6, pp. 845-
852.
Berenger, NL (1994). A Perfectly Matched Layer for the Absorption of Electromagnetic

Waves, J. Comput. Phys., Vol. 114, pp. 185-200.
Cerretelli and Biffi Gentili, IT (2009). Accuracy in Modeling Conformal Window Glass
Automotive Antennas by EM Numerical Solvers, 39th European Microwave Conference
(EUMC-2009), Roma, Italy.
Cerretelli and Biffi Gentili, IT (2007). Progress in Compact Multifunction Automotive

Antennas, International Conference on Electromagnetics in Advanced Applications
(ICEAA 07).
Engheta, Murphy, Rokhlin and Vassiliou, USA (1992). The Fast Multipole Method
(FMM) for Electromagnetic Scattering Problems, IEEE Trans. Antennas and Propagation,
Vol. 40, pp. 634-641.
Jakobus, Berger and Landstorfer, CH (2000). Efficient Techniques for Modelling

Integrated Windscreen Antennas within the Method of Moments, Millennium Conference
on Antennas and Propagation (AP 2000), pp. 102-105.
Krietenstein, Schuhmann, Thoma and Weiland, USA (1998). The Perfect Boundary
Approximation Technique Facing the Big Challenge of High Precision Field

Computation,, Proc. of the XIX International Linear Accelerator Conference (LINAC 98),
pp. 860-862.
Low, Langley and Batchleor, UK (2007). Modelling and Performance of Conformal

Automotive Antennas, Microwave Antennas & Propagation, IET, Vol. 1, pp. 973-979.
Low, Langley, Breden and Callaghan, USA (2006). Hidden Automotive Antenna
Performance and Simulation, IEEE Trans. Antennas and Propagation, Vol. 54, pp.3707-
3712.
Mur, USA (1981). Absorbing Boundary Conditions for the Finite-Difference

Approximation of the Time-Domain Electromagnetic-Field Equations, IEEE Trans.
Electromagn. Compat., Vol. EMC-23, pp. 377-382.
Nishikawa, JP (2003). Land Vehicle Antennas, IEICE Trans. Commun., Vol. E86-B,
No.3, pp. 993-1004.
Rao, Wilton and Glisson, USA (1982). Electromagnetic Scattering by Surfaces of

Arbitrary Shape, IEEE Trans. Antennas and Propagation, Vol. 30, pp. 409-418.
Song, Lu and Chew, USA (1997). Multilevel Fast Multipole Algorithm for
Electromagnetic Scattering by Large Complex Objects, IEEE Trans. Antennas and
Propagation, Vol. 45, pp. 1488-1493.
Taflove and Brodwin, USA (1975). Numerical Solution of Steady-State Electromagnetic

Scattering Problems using the Time-Dependent Maxwells Equations, IEEE Trans.
Microwave Theory Tech., vol. MTT-23, pp. 623.
von Hippel, Ed., USA (1954). Dielectric Materials and Applications, The M.I.T. Press,
Cambridge, Massachusetts.
Weiland, NL (1977). A Discretization Method for the Solution of Maxwell's Equations for
Six-Component Fields, Electronics and Communication, (AE), Vol. 31, p.116.

Section 2
Passenger and Transportation Safety and Services
Spread Spectrum Radar Technology for Automotive Applications: State of the Art
and New Proposals
Monitoring and Controlling Product and Service Engineering: The Role of the
Vehicles Repair Solutions
VHDL-AMS Simulation of an AHRS Platform for Automotive

Spread Spectrum Radar Technology for Automotive
Applications: State of the Art and New Proposals
Susanna Spinsante, Ennio Gambi
ABSTRACT
The development of effective anti-collision warning radars is a key step in the

technological initiatives promoted to improve road safety, and efficiency of transport
systems. The ultimate objective remains crash prevention, through an accurate knowledge
of the location, speed, acceleration or deceleration of the nearby vehicles. Automotive
radars provide an automatic vision of the environment where the vehicle is moving, to
draw the information required for guaranteeing, as much as possible, safety for the
vehicle and the passengers on board. About the expected performance, the target of an
automotive radar is to have a high detection capability but also a low false detection rate;
the analysis shown in this Chapter will be mainly focused on such figures.
Traditionally, Short Range Radars (SRR) and Long Range Radars (LRR) have been used.
Classic LRR solutions employ Frequency Modulation - Continuous Wave transmission,
that is very simple to implement but also prone to interfering signals produced by
neighboring radars of the same type. To overcome this limitation, radars employing
Spread Spectrum methods in Direct Sequence configuration have been introduced. All
spread spectrum systems make use of pseudorandom codes, that determine the frequency
spectrum occupied by the output signal, and control the spreading pattern of the system.
The accuracy in the distance evaluation (that is of course an important result of radar
detection) depends on the auto-correlation properties of the spreading sequences
associated to each radar equipment; on the other hand, the ability to reject the
interference due the other users depends on the cross-correlation properties of the whole
set of sequences adopted. Another important aspect concerns the number of signatures
available; in traditional techniques, this number depends on the sequence length, and the
latter cannot be arbitrarily large, because of the constraints on the bandwidth. To improve
these features and overcome existing limitations, the adoption of innovative sets of
spreading sequences, such as chaotic and De Bruijn sequences, is investigated and
evaluated as an alternative to more conventional solutions. This Chapter provides a brief
introduction to Spread Spectrum radar applications in the automotive scenario, by
highlighting the evolution from traditional to more advanced solutions, with a specific
focus on the selection of innovative sets of spreading codes, and the theoretical
evaluation of the performance obtainable.

INTRODUCTION
Chaos theory, a branch of the theory about nonlinear systems, has been intensively
studied in the past decades. Initially, it was investigated by researchers with strong
mathematical background, rather than circuit-designers or electronic engineers. This was
mainly due to the fact that circuit design and implementation could not match up with the
mathematical equations describing chaotic systems, because of technical and practical
implementation problems. With the advances in circuit technology and digital signal
processing, the exploitation of chaotic phenomena for a number of telecommunication
applications (Sobhy and Shehata, 2001), and in daily real-life engineering products,
became possible. Nowadays, more and more systems and applications may benefit from
using the chaotic dynamic behavior.
A chaotic system is particularly sensitive to environment changes and highly dependent
on the initial conditions settings. Even a small difference in the initial conditions
produces a very different chaotic signal, after a short time period. Therefore, it is possible
to generate a large number of chaotic signals, even with a very simple dynamic
deterministic equation.
Such a behavior makes chaotic systems highly unpredictable, and more similar to an ideal
white noise source than any other pseudo-noise (PN) source currently in use. This
resemblance to white noise and, more generally, to uncorrelated sources, reflects on the
very favorable auto- and cross-correlation properties that chaotic signals exhibit when
used as spreading sequences, in spread spectrum (SS) systems. Such correlation
properties can make these signals preferable against competing solutions, like those based
on Gold codes (Sivanesan and Beaulieu, 2001). Moreover, as said above, the number of
chaotic signals, for a prefixed bandwidth occupation and correlation properties, can be
very large, thus removing the problem on the number of codes available.
Spread spectrum techniques may be defined as methods by which the signal energy
generated in a particular bandwidth is deliberately spread in the frequency domain, thus
resulting in a signal with a wider bandwidth. Such techniques have become more and
more popular, especially as the available frequency spectrum is becoming more and more
crammed. Although spread spectrum systems are more complex and expensive to
implement than conventional systems, they have many advantages. The major one is the
inherent signal security they can provide. Unlike conventional systems, it is extremely
difficult to eavesdrop on a conversation that takes place over a spread spectrum system.
Conventional systems try to allocate as much information into as small a bandwidth as
possible. These systems, however, can easily be jammed by a high-power jamming signal
at a frequency that covers the frequency band of the particular system. They also output a
relatively high power to the antenna.
Spread spectrum systems, on the contrary, spread the signal over as wide a bandwidth as
possible; they also try to hide the transmitted signal as close to the background noise as
possible. This makes the communication very difficult to intercept, in the frequency
domain: it cannot be easily tracked, and it is more difficult to jam.
All spread spectrum systems make use of pseudorandom code generators. Such codes are
referred to as pseudorandom noise codes (PN codes). Each code in a given set is used to

set the frequency spectrum that the output signal will occupy. It also determines and
controls the spreading pattern of the system: by using a carefully designed and selected
spreading sequence set, which benefits from good correlation properties, it is possible to
implement multiple access technologies, meaning multi-user transmission at the same
time, in the same frequency band.
Chaotic signals are highly unpredictable, similar to an ideal white noise source, and very
easy to generate and control. The adoption of chaotic radars for automotive applications
has been already suggested (Sobhy and Shehata, 2001; Sivanesan and Beaulieu, 2001;
Gambi, 2008); however, the results presented were quite preliminary. In this Chapter, a
further contribution is discussed, through theoretical and numerical analysis, by
investigating the performance, in terms of detection capability and false detection rates,
of Direct Sequence Spread Spectrum (DS-SS) radars based on a class of chaotic
sequences. A comparison with a solution based on classical Gold codes is developed, by
considering typical, although simplified, road environments. The improvement that is
achievable by using chaotic sequences is well appreciated when focusing on the worst
operation conditions, with the automotive radar subject to blinding from other radars.
Under these very severe conditions, chaotic sequences can permit to increase the correct
detection probability (CDP), and to reduce the false detection probability (FDP).
Automotive Radar Systems
Improving road safety has become one of the most important priorities for any
government policy all over the world. Just to mention the position of the European
Commission, that inspires the policies of most European countries, a fundamental
objective is to improve safety, security, comfort and efficiency in all transport models
(...) focusing on advanced pilot/driver assistance systems, in support of vision, alertness,
manoeuvering, automated driving compliance with the regulations, etc. According to
this orientation, many research programs, like e-Safety (ECC, 2003), have been
developed, seeing the participation of academic and industrial partners. The situation is
not different outside Europe (SAFETEA-LU, 2007; ASV, 2007). In the USA, the
National Highway Traffic Safety Administration (NHTSA) has reported that there are 6.2
million crashes annually resulting in more than 43,000 fatalities and a cost to society of
more than $230 billion (Wenger, 1998). Half of these fatalities occur in cases of vehicles
leaving a road and passing through intersections. In addition, injuries and damages from
non-fatal accidents lead to significant costs in terms of healthcare and property.
Safety systems traditionally applied by automakers are based on individual vehicle
implementations, such as airbags and anti-lock brakes. However, despite some substantial
initial improvements to road safety provided by these solutions, single-vehicle systems
have not been able to further reduce fatalities or injuries significantly. The number of
death events has kept gradually increasing in recent years, even though most of the
nations have strong laws and policies supporting highway safety. Road accidents are
often caused by driver carelessness or ignorance, simple misconduct, or lack of
experience. In some instances, dangers due to severe weather and road conditions, and
consequent obstruction of view, are also responsible.

It is extremely difficult to eliminate these human factors due to the inherent limits of
human sensing and reaction capabilities; most people all over the world give their lives in
transportation related accidents, more than any other single cause of death. In this
context, the Wireless Access in Vehicular Environments (WAVE) technology is a
revolutionary solution for vehicle safety enhancement, providing drivers with early
warning, perceive and assistance. It is an extension of human natural sensing, and enables
tele-sensing of vehicles. Working as probes, vehicles report timely traffic and road
condition information to transportation agencies; the same information is thereafter
shared by a large community of users. The standardization activity of the WAVE system,
through the initiatives of the Intelligent Transportation System (ITS) Committee of the
IEEEs Vehicular Technology Society, that led to several recommendations published
from 2006, introduces the next generation, dedicated, short-range communications
technology, which provides high-speed vehicle-to-vehicle and vehicle-to-infrastructure
data transmission, and has major applications in ITS, vehicle safety services, and Internet
access.
The final WAVE architecture provides specific protocols for real-time safety message
delivery in the vehicular environment, to effectively support road safety. Once in place
(cars with WAVE started come off the assembly line in 2011), the system will provide
driver alerts, including visual, audible, and tactile warnings. A warning of an impending
accident will be fed to pre-crash systems, like those now found only on some luxury cars,
to perform specific actions, like pre-tensing seat belts, prepare brakes for an emergency
stop, and tilt reclined seats upright. However, the most important objective remains to
prevent the crash through an accurate knowledge of the location, speed, acceleration or
deceleration of the surrounding vehicles. The auto-industry is pouring a lot of efforts and
money in these projects, and the expectation level is very high.
Automotive radars permit an automatic vision of the environment where the vehicle is
moving, and, from that, to draw the information required to perform some actions that
guarantee, as much as possible, safety for the vehicle and the passengers onboard. In
designing radar systems for automotive applications, it is necessary to comply with
national constraints on the usable frequencies and allowed power levels. In particular,
several recommendations have been already issued by national and international agencies
(ETSI, 2004; ETSI, 2005; FCC, 2006; ETSI, 2004), that put strict limits on the power and
the spectral occupancy allowed to radar equipments. These are necessary for
electromagnetic compatibility with other systems, and to ensure the absence of any effect
on the human health. Moreover, it is mandatory to limit the costs: only cost-efficient
systems will be accepted by the market for a widespread use, which in turn is a
prerequisite for effective public benefit. Another important technological and commercial
requirement concerns the need to keep small sensor sizes, due to their location in the
vehicle frame.
From the performance viewpoint, the target of an automotive radar is to have a high
detection capability but also a low false detection rate; however, other issues, like
installation, maintenance and auto-calibration, must be taken under control as well.

Two kinds of radar have been considered for automotive applications up to now: Short
Range Radars (SRR) and Long Range Radars (LRR). In the framework of European
regulations, SRRs operate in the ISM band of 24 GHz or in the 77 GHz to 81 GHz band,
with a visibility range from 0 to 30 meters, approximately. They are used for a number of
applications, to enhance the active and passive safety for all kind of road users.
Applications that enhance passive safety include obstacle detection, collision warning,
lane departure warning, lane change aid, spot detection, parking aid and airbag arming, as
depicted in Fig.1. The grouping of these functions is referred to in the literature as a
safety belt for cars. The SRR is a combination of two functions: it allows a precise
speed measurement combined with an accurate radial range information. To obtain the
required resolution, in the order of approximately 5 to 10 centimeters, the SRR needs a
large bandwidth of 4 GHz, for the range measurement.
LRRs operate between 76 GHz and 77 GHz with a maximum bandwidth of 1 GHz. They
allow an operating range of approximately 150 meters and are used at vehicle velocities
not below 30 Km/h. One or multiple narrow lobes control and scan the driving path in
front of the car, to determine the distance to the vehicle running ahead, thus maintaining a
constant minimum safety gap, and allowing an Adaptive Cruise Control (ACC).
Fig.1. Possible applications for automotive radars.
Limiting our focus on LRRs, classic implementations employ signal processing

techniques based on Frequency Modulation-Continuous Wave (FM-CW) transmission,
and a 1 GHz bandwidth at the most. FM-CW systems are very simple to implement but
their performance can be strongly degraded because of the interfering signals produced
by neighboring radars of the same type. To overcome this limitation, radars employing
Spread Spectrum Pseudo Noise methods in Direct Sequence configuration have been
proposed.
The specific PN sequence is the radar signature, which allows to compute the distance
of a vehicle by selecting its echo among a very large number of echoes due to other
vehicles, but also to the signals coming from other radars operating in the same frequency
bandwidth. The accuracy in the distance evaluation depends on the auto-correlation
properties of each sequence; the ability to reject the interference due to the other users

depends on the cross-correlation properties of the whole set of sequences adopted.
Another important aspect concerns the number of signatures available; in case of PN
sequences, this depends on the sequence length, and there are relatively small sets of
signatures available. For these reasons, it seems meaningful to explore possible
applicability of chaotic signals in place of more conventional solutions, to increase the
number of available signatures and maintain, at the same time, the required correlation
properties.
This Chapter provides insights about the possibility of adopting chaotic sequences as
radar signatures in automotive applications, to overcome the limits of PN sequences
discussed above. Some preliminary evaluations are also provided for an alternative set of
spreading codes, i.e. the De Bruijn binary sequences, that up to now have been mainly
applied in cryptographic contexts. The performance obtainable are shown and discussed,
in different application scenarios; open issues and problems are also evidenced.
SPREADING SEQUENCES FOR DS-SS SYSTEMS
In this section, a brief overview of the sequences considered as spreading codes is

provided.
Gold Sequences
Binary Gold sequences are a well known set of PN sequences, also used in wireless
mobile communications (WCDMA) as scrambling codes, either in uplink (for separating
the mobile stations) or in downlink (for separating the cells). They are obtained by
combining two maximum-length shift register sequences (M-sequences) (Proakis, 1995)
of length N = 2m - 1, where m is the size of the shift-register. The Linear Feedback Shift
Registers (LFSR) and the connections selected to generate m-sequences of length 32, 64
and 127, respectively, are illustrated in Fig. 2.
Gold sequences exhibit a three-valued cross-correlation function, with values {-1, -t(m),
t(m) - 2}, where t(m) is equal to:
2 ( m1) / 2 1
t ( m) ( m 2 ) / 2 (for odd and even m respectively) (1)
2 1
The characteristic three-valued cross-correlation function of a pair of Gold sequences of
length 63 is shown in Fig.3.

Fig.2. Gold sequences generators.
Fig.3. Cross-correlation function of a pair of Gold sequences of length 63.
Also the auto-correlation function is a three-valued function, with values {N, -t(m), t(m) -
2}, as illustrated in Fig.4.
The number of sequences, for a given N, is N + 2. Since the sequence length equals the
Spreading Factor (SF), for an assigned SF, the number of spreading sequences is SF + 2.
Gold codes show good auto- and cross-correlation properties, particularly for large values
of m. These properties, however, can be outperformed using chaotic sequences, whose
rationale is discussed in the following.

Fig.4. Auto-correlation function of Gold sequences of length 63.
Chaotic Sequences
Chaotic signals have been recently valued in CDMA systems, mostly because of their
favorable correlation properties (Rovatti, 2004; Chiaraluce, 2002; Rovatti et al., 2004).
The perfect random generator would be an ideal white noise source. A number of
physical processes, relying on phenomena that produce random noise at a microscopic
scale, could be used, in principle, to mimic a completely uncorrelated random source.
Anyway, these natural phenomena are generally hard to control and/or to model by a
mathematical approach. For this reason, physical sources are often replaced by pseudo-
noise sources that, being necessarily based on finite memory algorithms, exhibit periodic
behaviors and generate correlated examples. Therefore, their behavior may be
significantly far from the ideal one. Gold sequences are examples of signals produced by
pseudo-noise sources.
A significant advance towards the idea of realizing an artificial source with features very
similar to the ideal ones is represented by chaotic generators. They have a deterministic
basis, like pseudo-noise sources; their peculiar feature is to exploit a simple basic rule, in
an iterative process, that rapidly produces a very complex and quite unpredictable
evolution. As a consequence, signals obtained by chaotic sources are very irregular, not
periodic, and extremely sensitive to the initial conditions: two identical systems,
apparently starting at identical conditions, will end up with totally different outputs.
These behaviors correspond to the commonly recognized concept of chaos and are very
appealing in the perspective of designing good spreading sequences.
Chaotic sources are obviously available also in nature, but the advantage of an artificial
apparatus is in the possibility to obtain the desired chaotic behavior through elementary
mathematical rules, that are quite easy to control (at least in the generation process) and
to implement by electronic circuits. As a matter of fact, a simple Bernoulli shift map, that
is a piece-wise affine transformation, can behave like an ideal binary random generator
and can be implemented with a single stage of a pipeline analogue-to-digital converter

(Callegari, 2005). In our study, we used the chaotic evolution resulting from the solutions
of the following, well-known, Lorenz system of equations:
du
(v u)
dt
dv
r u v 20 u w (2)
dt
dw
5u v b w
dt
In the system, , r and b are suitably fixed parameters (a typical choice assumes = 16, r
= 45.6 and b = 4) while u(t), v(t) and w(t) are the system states: each state variable can be
used for providing the time evolution of a chaotic spreading signal. The Lorenz system is
solved starting from assigned initial conditions (for example u(0) = 0.82, v(0) = 0.63 and
w(0) = 0.74); the result is an unpredictable set of not periodic functions, each one
showing features very similar to white noise and therefore with excellent correlation
properties. The time evolution of the u(t) state, with the following parameters, is shown
in Fig.5.
16 u 0 0.82

r 45.6 v 0 0.63

b4 w 0 0.74 (3)
For a given set of initial conditions, different chaotic spreading signals can be obtained
by generating a long chaotic evolution and then braking it into a desired number of
segments (each segment represents a single spreading signal). This is the process applied
in our study. Alternatively, different spreading signals may be generated by solving the
Lorenz system with different initial conditions.
Fig.5. Time evolution of the u(t) state.
The solutions to the Lorenz system are real functions. In this sense, they can be seen as
multilevel signals. Their amplitudes can be quantized in order to obtain a fully digital
sequence. Sometimes, chaotic sequences are obtained directly in binary form (Callegari,

2005; Lon, 2004; Biondi, 2004). Such an approach may provide advantages in the
implementation, by smoothing some practical difficulties about the repeatability of
chaotic maps. Anyway, the conclusions drawn for Lorenz signals can be considered
valid, with a good approximation, for the binary chaotic sequences too. An interesting
property of chaotic generation is the possibility to increase the number of signals with
specified correlation properties. This feature is useful in the automotive context, where
the need of correctly discriminating a (possibly very) large number of users is a
distinctive property.
A means for specifying the desired correlation properties of the set of spreading codes
relies in fixing suitable thresholds for their auto- and cross-correlation (AC, CC) values.
Noting by cnj the n-th chip of the j-th spreading sequence, the correlation at a shift
between the i-th and the j-th sequences is defined as:
N 1
Pij ( ) c ni c nj
n 0 (4)
where N is the sequence length. The normalized maximum autocorrelation value, for
0 results in:
1
PA max P jj ( )
0 P jj (0)
(5)
whilst the maximum cross-correlation between the j-th sequence and all the other
sequences in the set is:
1
PC max Pij ( ) i [1..S ]
P jj (0) i j
(6)
Following (Morantes, 1998), from a set of S chaotic signals randomly generated, a first
selection is made by eliminating those sequences with P A greater than a threshold 1;
then, if applicable, the selection is refined by eliminating those sequences with P C greater
than a threshold 2. An example based on the first selection strategy, with a variable 1, is
shown in Table 1.
Table 1. Number of selected sequences fixing a threshold on the auto-correlation.
1 Number of sequences
0.62 400
0.61 0.55 400 396
0.545 0.485 395 382
0.48 0.42 380 337
0.419 0.355 335 229
0.354 0.291 229 107
0.29 0.226 107 19
0.225 0.2 18 5

The starting set contains S = 400 chaotic sequences, with N = 31. All the sequences have
PA smaller than 1 = 0.62, but 229 out of them, for example, have P A smaller than 1 =
0.355; only 107 sequences exhibit a PA smaller than 1 = 0.291. When the second
selection criterion is applied, the number of sequences is, obviously, further reduced. An
example is reported in Table 2, where the same sequences with P A 0.291, selected
before, have been tested for their cross-correlation value. All the 107 sequences have P C
smaller than 2 = 0.81, but only 7 out of them, for instance, exhibit a P C 0.55. It should
be noted, however, that such a result depends on the relatively small size of the starting
set (S = 400); by increasing the size of the starting set, larger amounts of chaotic
spreading sequences with prefixed correlation properties should be found.
Table 2. Number of selected sequences fixing a threshold on the cross-correlation (1 =

0.291).
2 Number of sequences
0.81 107
0.8 0.745 107 106
0.74 0.678 106 92
0.677 0.613 92 41
0.612 0.55 41 7
0.545 0.485 73
0.48 0.42 31
0.415 1
In the case of Lorenz chaotic signals, the spreading factor does not result naturally
from the solution of the Lorenz system. For a given set of parameters and initial
conditions, signal evolution depends on the discretization step h (for computational
purposes, the Lorenz system is in fact approximated by an equivalent discrete form) and
on the sampling period Tk:
u t 1 u t h ( v t u t )

v t 1 v t h ( r u t v t 20 u t w t )
w w h (5 u v b w )
t 1 t t t t
(7)
A possible procedure consists in fixing the value of Tk and then assuming a number of
chaotic samples that ensure (approximately) the same bandwidth expansion factor of the
competing Gold sequences. Such a strategy, that implies to develop a parallel analysis in
the frequency domain, seems also the most proper one for ensuring a fair performance
comparison. Even more important, the selected chaotic evolutions exhibit auto- and
cross-correlations with reduced dispersion around the average value: at a parity of the
number of sequences, the auto- and cross-correlations of chaotic signals have, typically,
smaller standard deviations (Chiaraluce, 2002). As performance depends on the variance
of the disturbance, it is reasonable to expect this may have a positive impact on the
performance results.

Binary De Bruijn Sequences
Binary De Bruijn sequences (Mayhew, 2000; Mayhew, 2007; Andrenacci, 2010) are
nonlinear shift register sequences with maximal period, N = 2 n: n is called the span of the
sequence, i.e. the number of stages in the generating shift register (Jansen, 1991). They
may be generated by n-stage nonlinear feedback shift registers, as they include the all
zero n-tuple, which cannot be included if a linear shift register is adopted for the
generation process. In the binary case, the total number of distinct sequences of span n is
( n 1)
n
22 . The states S0, S1, . . . , SN1 of a span n De Bruijn sequence are exactly the 2n
different binary n-tuples; when viewed cyclically, a De Bruijn sequence of length 2 n
contains each binary n-tuple exactly once over a period. Let us consider the simple case
of n = 3. For n = 3, two distinct De Bruijn sequences exist. Each sequence contains all the
possible binary 3-tuples (000, 001, 010, 011, 100, 101, 110, 111) exactly once over a
period of length 23 = 8. The sequences are:
s1 = 00010111
s2 = 11101000
De Bruijn sequences have many desirable properties, such as long period and low
predictability, and can be used as stream ciphers in cryptographic applications. As
discussed by Mayhew in (Spinsante, 2011), a De Bruijn sequence may also be obtained
by adding a single zero to the longest run of zeros in an M-sequence, thus obtaining a so
called primitive De Bruijn sequence; at the same time, a modified De Bruijn sequence is
obtained when removing a zero from the all zero n-tuple of a De Bruijn one. However,
not all the modified De Bruijn sequences are M-sequences: the latter represent the linear
subset of the modified De Bruijn sequences family.
Being De Bruijn sequences maximal period sequences, their length is always an even
number, and a power of 2 in the binary case. When comparing the cardinality of a set of
binary De Bruijn sequences, for a given value of the span n, to the cardinality of other
families of sequences of corresponding length (such as Gold or M-sequences), it is clear
that, at least theoretically, the exponential law in the De Bruijn case could make it
possible to enlarge the set of potential users in a spread spectrum system, in case also the
other necessary properties are verified. As an example, for n=10 there exist up to 60
binary M-sequences of length 1023, up to 1025 binary Gold sequences of length 1023,
and up to 2502 binary De Bruijn sequences, that is an extremely huge amount. Of course,
not all the 2502 sequences may anyway be suitable for application in a multi-user system.
In any case, even if strict selection criteria are applied, it is reasonable to expect that a
quite huge amount of sequences may be extracted from the entire family.
Several results are known about the correlation properties of binary De Bruijn sequences.
The autocorrelation PA (k ) of a binary De Bruijn sequence c = (c0, c1, , cL-1) of length
L = 2n, for a given shift k, may assume only a set of given values:

PA (k ) 2 n , for k 0
PA (k ) 0, for 1 | k | n - 1 (8)
PA (k ) 0, for | k | n
The second property implies that as long as the span of the sequence increases, there are
more null samples of the autocorrelation sidelobes (i.e. the values assumed by
PA (k )
for k 0) that provide a better peak isolation. It is also known that

PA (k ) 0 mod 4
for all k, for any binary sequence of period L = 2n , with n 2.
Provided that in a binary De Bruijn sequence the number of 1s equals the number of 0s,
it turns out that any binary De Bruijn sequence, when converted into a bipolar form, has
a zero average autocorrelation. Fig.6 shows the average autocorrelation profile computed
over all the 2048 binary De Bruijn sequences of length 32. With the exception of the peak
value at k = 0, the autocorrelation profile is symmetric with respect to the central value of
the shift, k = 16. As a matter of fact, it is possible to prove that
PA (k ) PA ( L k ), for 0 k L - 1 . So, when n increases, the autocorrelation
profiles of the De Bruijn sequences will show many samples equal to 0, a symmetric
distribution of the samples, and a reduced number of different positive and negative
samples, as to give an average autocorrelation equal to 0.
Fig.6. Average autocorrelation profile of binary de Bruijn sequences of length 32
Fig.7 shows the set of autocorrelation profiles of all the 2048 binary De Bruijn sequences
of length 32, with the exception of the peak value (that is obviously 32 in all the cases), to
improve figure readability. The symmetry of each graph and the presence of null values
are easily recognized. The first (n 1) samples adjacent to the peak, that is located at k =
0, are always equal to 0, as discussed above. This is an important feature to improve the

detection capability of the autocorrelation peak, i.e. the matched filtering of the received
sequence.
Fig.7. Autocorrelation profiles of all the 2048 binary de Bruijn sequences of length 32
(peaks removed)
For the cross-correlation function of a pair of De Bruijn sequences a and b (a b) of the

same span n and length L, the following bound holds:
2 n PC ( k ) 2 n 4, for 0 k L - 1 , where PC(k) denotes the cross-correlation
between the two sequences. Any couple of sequences including a De Bruijn sequence c
and its complementary one c*, that is also a De Bruijn sequence and belongs to the same
family, may provide a negative peak of cross-correlation equal to -2n for a specific value
of the shift k, that motivates the lower bound on the cross-correlation profile. If we need
to avoid this negative peak value of the cross-correlation, only half of the sequences in
the set should be considered, by taking just one sequence in each couple (c,c*). The
cross-correlation function of binary De Bruijn sequences shows symmetry properties
similar to those exhibited by the autocorrelation function. All the possible cross-
correlation values are integer multiple of 4.
DS SS CHAOTIC RADAR: ANALYSIS AND SIMULATION RESULTS
This section presents the peculiarities of the DS-SS radar and an algorithm developed to
improve the related detection capability. Performance in terms of correct detection
probability and false detection probability are discussed, by means of numerical
simulations.
DS/SS Radar Basics
A DS/SS radar is basically an electromagnetic system for the detection and location of
obstacles or targets, that operates by radiating energy into space and detecting the echo

signal reflected from an obstacle, or target, in the detection area. In a DS/SS
configuration, each radar uses a different spreading sequence, composed by N chips, that
modulates the transmitted signal.
Range to a Target
The range to a target is determined by the time TR it takes the radar signal to travel to the
target and back. Electromagnetic energy in free space travels with the speed of light,
which is c = 3 x 108 m/s. Thus, the time for the signal to travel to a target located at a
range R and get back to the radar is 2R/c. The range to a target is then:
cT
R R
2 (9)
With the range expressed in Kilometers, and TR in microseconds, the range can be written
as R ( km ) 0.15 TR ( s ) : in other words, each microsecond of round-trip travel
corresponds to a distance of 150 meters. Therefore, in order to calculate the target
distance R, it is necessary to measure TR, which is evaluated in a DS radar through the
correlation between the received spreading code and the reference spreading code.
Maximum Unambiguous Range
Once a spreading sequence is radiated into the space by a radar, a sufficient time must
elapse to allow all echo signals get back to the radar receiver, before the next spreading
sequence is transmitted. The rate at which sequences may be transmitted, therefore, is
determined by the longest range at which targets are expected. In a LRR the maximum
detectable range Rmax is 150 meters, which corresponds to a maximum round-trip travel
of one microsecond, as illustrated above. Thus, the transmitted signal will be modulated
by a spreading sequence, periodically repeated, lasting one microsecond, in order to
detect a vehicle without ambiguity.
By assuming that each sequence has a duration T = 1 s and length N, the chip duration
defined as:
T
TC
N (10)
will determine the range resolution R as:
c TC
R
2 (11)
The Radar Equation
The radar equation relates the range of a radar to the characteristics of the transmitter,
receiver, antenna, target, and the environment. If the transmitter power Pt is radiated by
an isotropic antenna, the power density at a distance R from the radar is equal to the
radiated power divided by the surface area 4R2 of an imaginary sphere of radius R:

Pt
Power density at range R from an isotropic antenna
4 R 2 (12)
Radars, however, employ directive antennas, with gain G, to concentrate the radiated
power in a particular direction. Therefore the power density can be written as:
Pt G
Power density at range R from a directive antenna
4 R 2 (13)
The target intercepts a portion of the incident energy and radiates it in various directions.
Only the power density re-radiated in the direction of the radar (the echo signal) is of
interest. The radar cross section of the target determines the power density returned to
the radar for a particular power density incident on the target; it is denoted by and is
often called RCS. The radar antenna captures a portion of the echo energy incident on it.
The received power is given as the product of the incident power density times the
effective area Ae of the receiving antenna. The received signal power Pr is then:
P GA
Pr t 2 e 4
(4 ) R (14)
The maximum range of a radar is the distance beyond which the target cannot be
detected. It occurs when the received signal power P r just equals the minimum detectable
signal.
By assuming that the RCS equals 1 for all the detected targets, and leaving Pt, G and Ae
out, since they are the same for all the echoes, the received signal power of the i-th echo
can be written as:
1
Pr ,i
(4 )2 R i 4 (15)
where Ri is distance of the i-th target, whereas the amplitude of the i-th signal Ar,i is
proportional to:
1
4 2R i 2 (16)
Proposal of a Detection Algorithm for a Multi-user Radar Environment
A major difficulty, in automotive radar operation, is caused by the presence of several

targets in the scanned region, which is responsible for the appearance of multiple echoes
as well. Let us consider Fig.8: the radar on board of the leftmost car intercepts three
vehicles, A, B and C, at distance LA = 40 m, LB = 90 m and LC = 140 m, respectively. By
assuming, for simplicity, a unitary cross section for each vehicle, the amplitude of each
signal detected by the radar sensor after the round trip path is proportional to 1/L i2, with i
= A, B, C. Then the received signal:
1 1 1
y( t ) 2
x(t A ) 2 x(t B ) 2 x(t C )
LA LB LC (17)

consists of the superposition of three replicas (echoes) of the transmitted signal x(t) but
with different delays i, that the sensor system aims to estimate. Fig.9 shows the AC
functions, for the example discussed, obtained by using Gold sequences of length NG =
63 and NG = 127. For better evidence, the results are plotted directly as a function of the
distance L. A peak is evident at LA = 40 m, which is due to (and permits to detect the
presence at this distance of) vehicle A. Correlation peaks for the other vehicles, instead,
are rather small (because of their larger distance), and apparently masked in the figure.
Fig.8. Example of multiple targets in the radar operating range.
Fig.9. Auto-correlation for the case in Fig.8.
On the other hand, the radar system should determine also the distance from the other
vehicles. Once detected the greatest peak, typically through the comparison with a
threshold, the processor on board of the vehicle can subtract the detected sequence, with
its own delay and amplitude, from the received signal, to eliminate the dominant
contribution and discover the others. A new correlation is then computed on the cleaned
signal, searching for a new AC peak. As an example, Fig.10 shows the second correlation
computed, which evidences, for both the considered sequence lengths, a peak at LB = 90
m, which is due to (and permits to detect the presence at this distance of) vehicle B. The
procedure herein described is iterated, until no peak above the threshold is found.
Another obvious criterion to stop the procedure consists in not accepting, from the
algorithm, peak positions that are at a shorter distance than those already found. It is
important to stress that signal processing required for these operations is very fast, so that
a large number of peaks can be detected in a very short time.
The position of the threshold should be optimized, as a function of the number and
distance of the vehicles that the system should detect at each scan; too much small

thresholds should be avoided, as they can cause a large number of false detections. On
the other hand, also the risk of false detection is strictly related to the correlation
properties of the sequences used, thus implying an in depth study.
Fig.10. Auto-correlation for the case in Fig.8, after elimination of the effect of the
dominant term.
Continuing the example in Fig.8, the algorithm has been tested over 20 Gold sequences
(for each length) and assuming a threshold equal to 2.5 times the correlation standard
deviation; in all cases, vehicles A, B and C have been detected, and their distance
correctly estimated. Then, Gold sequences seem adequate to solve the problem of
intercepting vehicles that are moving along the same direction of the tagged car.
Actually, in this preliminary example, we have neglected the effect of the interfering
signal produced by other radar equipped vehicles, whose echoes may disturb reception.
For example, vehicle A in Fig.8 may be also equipped with a radar: its signal is then
reflected from vehicle B and adds to the total signal received by the tagged sensor.
Through simulation, we verified that 6 of the selected Gold sequences with length N G =
127 become unable to detect the presence of vehicle B because of the interfering signal.
The impact of an interfering signal is, obviously, much stronger when originating from a
vehicle that moves in the opposite direction (from right to left, in Fig.8). An example is
shown in Fig.11. The interfering radar, at distance LI, obviously uses a spreading
sequence different from that of the radar under test. However, the amplitude of the
interfering signal is proportional to 1/LI , and a serious risk exists that it masks all the
other signals. In practice, although the interfering signal weakly cross-correlates with the
useful emitted signal, the more favorable attenuation law can blind the radar. This
occurs when the ratio between the useful signal and the interfering signal becomes
smaller than a minimum value (for example, Signalto-Interference Ratio, SIR < 0.03).

Fig.11. Possible scenario for DS/SS radar operation.
It should be noticed that such very unfavorable situations take place only when the
interfering signal is received in the direction of maximum antenna gain. Because of the
relative movement of the antennas, whose beam is electronically controlled, this happens
only for a fraction of time in a vehicular scenario. Simple systems can be designed, for
example based on iterated measurements, which are able to overcome, in real time, the
criticalities. On the other hand, for larger SIR values, the ability to detect a vehicle by a
single measurement depends on the correlation properties of the spreading signal
adopted. Numerical examples will be provided, after having introduced possible usage of
chaotic signals, thus realizing a comparison between the two solutions.
Besides the probability of correct detection, we have to consider the probability of false
detection, i.e. the radar detects an obstacle that actually is not present. A false detection
occurs when a correlation value greater than the fixed threshold does not correspond to a
vehicle echo but, instead, to an interfering signal generated by another radar. So, too
much small thresholds should be avoided, as they can cause a large number of false
detections. The threshold required to minimize false detections depends on the SIR and
the sequence length adopted. For example, simulations show that for any SIR 0.03, the
probability of false detection is almost zero by assuming a threshold 3 times larger than
the AC standard deviation, when using Gold sequences of length NG = 63. A threshold
2.5 times larger than the AC standard deviation, when using Gold sequences of length NG
= 127, is needed. It is clear that the selection of the most suitable threshold value is a
trade-off between the need to have a high probability of correct detection (that increases
with threshold reduction), and a low probability of false detection (that shows an opposite
behavior).
Performance Improvement by Using Chaotic Spreading Signals
The maximization of correct detection probability, and minimization of false detection

probability, are conditioned on the adoption of spreading sequences with good correlation
properties. Gold codes can ensure acceptable performance, but the correlation properties,
and other features of the system, can be improved by replacing Gold signals with chaotic
signals.
In this work, focus is on the use of the chaotic evolution resulting from the solutions to
the Lorenz system (see Eq.2). To compare the performance of different spreading
sequences, an important requirement concerns the need to assume almost the same

bandwidth expansion factor, otherwise the comparison would be unfair. Bandwidth is
determined on a power content basis, i.e. as the frequency range including a given
percentage (95% for example) of the total power. In the case of Lorenz chaotic signals,
the bandwidth factor does not result naturally from the solution of the system. The
system is approximated by an equivalent discrete form:
u t 1 u t h ( v t u t )

v t 1 v t h ( r u t v t 20 u t w t )
w w h (5 u v b w )
t 1 t t t t
(18)
for a given set of parameters and initial conditions, signal evolution depends on the
discretization step, and on the sampling period. A possible procedure consists in fixing
the sequence duration T, and then assuming a number of chaotic samples that ensure
(approximately) the same bandwidth expansion factor of the competing Gold sequence.
Therefore, the number of samples in a chaotic sequence derived from the Lorenz system,
NL, has the same meaning of the length NG of a Gold sequence. Maintaining the duration
T and extracting, at a given rate, NL = 124, 252 and 508 chaotic samples, we have almost
the same 95% power bandwidth occupancy of Gold sequences with N G = 31, 63 and 127,
respectively. About the dispersion of the CC and AC values around the mean (i.e., the
standard deviation), we have calculated the AC and CC standard deviation for 8 Gold
sequences and 8 chaotic segments, at a parity of bandwidth occupancy. The various cases
have been denoted by Gold N and Lorenz N respectively. The results in Table 3 show
that the values for the chaotic segments are always smaller.
Table 3. CC and AC standard deviation for a set of 8 Gold sequences and 8 chaotic
segments at a parity of the bandwidth occupancy.
CC Standard Deviation AC Standard Deviation

Gold 31 0.178 0.177
Lorenz 124 0.090 0.083
Gold 63 0.123 0.125
Lorenz 252 0.063 0.059
Gold 127 0.087 0.088
Lorenz 508 0.044 0.043
Generally speaking, it is possible to verify that the distribution of the AC and CC values
is more similar to a Gaussian distribution with zero mean. Since the interfering effect is a
decreasing function of the standard deviation, and is more favorable in the case of a
Gaussian-like distribution, it is easy to think that the adoption of the chaotic solution in
the DS-SS radar should enable more accurate and robust detections.
The potential advantages offered by chaotic sequences are manifold. First of all, chaotic
sequences exhibit NL > NG; equivalently, the chip duration TC of the spreading
sequence is smaller. As in a radar system the value of T C determines the resolution R,
according with Eq. 11, by using chaotic codes it is possible to have more precise distance

estimations. As shown in Table 4, by using the chaotic codes, a four times smaller range
resolution is achieved.
Table 4. Range resolution.
Gold 31 Gold 63 Gold 127 Lorenz 124 Lorenz 252 Lorenz 508
R 4.84 m 2.38 m 1.18 m 1.21 m 0.6 m 0.3 m
The most important benefit of using chaotic signals in place of Gold sequences is in the
possibility to increase the probability of correct detection and to reduce the probability of
false detection, at a parity of the other parameters. A first example is given in Table 5: for
a scenario of the type reported in Fig.11, where an interfering radar is present, we
computed the percentage of correct detections for the case of Gold sequences and chaotic
(Lorenz) sequences, with different length, and by choosing different threshold values
(Thr*standard deviation of the computed correlation), to detect the correlation peaks.
Results refer to the most unfavorable condition, when the interfering radar is emitting its
signal exactly in the direction of the receiving antenna maximum gain. Besides the codes
mentioned above, a chaotic sequence with NL = 1020 is also considered, that, as
expected, exhibits the best performance. From Table 5 we see that, at a parity of the
bandwidth expansion factor, the chaotic signals generally ensure higher percentages of
correct detection. On the other hand, the detection ability improves for increasing
bandwidth expansion; in particular, the chaotic signal with N L = 1020 ensures 100% of
successful detection, even for a SIR value as low as 0.23. Conversely, for SIR = 0.03, the
radar is totally blind, independently of the spreading sequence adopted; as discussed
previously, however, this out-of-service condition is limited to a very short time,
because of the vehicle/antenna movements, and can be compensated through suitable
signal processing procedures.
Besides the probability of correct detection, mentioned above, also the probability of
false detection shall be taken into account, i.e. the probability that the radar detects an
obstacle in the scanned area which is actually not present. In particular, a false detection
is generated when there is a correlation peak above the threshold that does not correspond
to a vehicle. Table 6 reports a number of examples in this sense. Results have been
obtained by repeating simulation with (a large number of) different pairs of spreading
sequences for the tagged and the interfering radars, and enumerating the number of cases
when false detections were revealed, versus the total number of simulations.
A too small threshold value should be avoided, since it can cause a large number of false
detections. Moreover, there is not a threshold value suitable for all kind of sequences, but
the choice is related to the type of sequences adopted, and it is a trade-off between the
need to have a high probability of correct detection, and a reduced occurrence of false
detections.

Table 5. Percentages of correct detections as a function of the SIR for Gold and Lorenz
signals.
Gold 63 Lorenz 252 Gold 127 Lorenz 508 Lorenz 1020

Thr
2.5 3 3 3.5 2 2.5 3 3.5 4 3.5 4
SIR
0.5 80% 80% 85% 100% 90% 100% 90% 100% 100% 100% 100%
0.25 10% 10% 50% 50% 70% 65% 90% 95% 95% 100% 100%
0.23 10% 10% 50% 45% 70% 25% 85% 95% 90% 100% 100%
0.17 10% 5% 30% 20% 15% 15% 70% 50% 45% 90% 80%
0.15 10% 0% 20% 5% 15% 15% 65% 50% 25% 80% 60%
0.125 10% 0% 20% 5% 15% 15% 40% 30% 25% 55% 45%
0.1 10% 0% 10% 5% 15% 10% 25% 25% 10% 35% 20%
0.07 0% 0% 5% 0% 15% 0% 15% 0% 0% 15% 5%
0.03 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%
Table 6. Percentages of false detection as a function of the SIR for Lorenz signals.
Lorenz 252 Lorenz 508 Lorenz 1020

Thr
3 3.5 3 3.5 4 3.5 4
SIR
0.5 0 0 5% 0 0 0 0
0.25 5% 0 15% 5% 0 0 0
0.23 5% 0 15% 5% 0 0 0
0.17 5% 0 25% 5% 0 0 0
0.15 10% 0 30% 5% 0 0 0
0.125 10% 0 40% 5% 0 0 0
0.1 10% 0 45% 5% 0 0 0
0.07 10% 0 50% 0 0 0 0
0.03 10% 0 50% 0 0 0 0
When adopting chaotic sequences of length N=508 for small values of the SIR, the
threshold that guarantees the higher correct detection percentages also determines the
higher rate of false detections, as shown again in Fig.12.
Table 7 shows the threshold values chosen for the simulations presented in the following
part of this chapter, as the best trade-off between the need to have a high probability of
correct detection, and a low occurrence of false detections.
Table 7. Optimized values to set the threshold.
Gold Lorenz Gold Lorenz Gold Lorenz Lorenz

31 124 63 252 127 508 1020
Threshold 2.5 3 3 3.5 2.5 3.5 3.5

Fig.12. Percentages of sequences providing false alarm (a) and percentages of correct
detection (b) as a function of the SIR for Lorenz 508 signals for different values of the
threshold.
Lorenz 508 (thr=3) Lorenz 508 (thr=3.5)

60%
50%
% sequences providing false alarms
40%
30%
20%
10%
0%
0,5 0,25 0,23 0,17 0,15 0,125 0,1 0,07 0,03
SIR
(a)
Lorenz 508 (thr=3) Lorenz 508 (thr=3.5)
100
90
80
70
% correct detection
60
50
40
30
20
10
0
0,5 0,25 0,23 0,17 0,15 0,125 0,1 0,07 0,03
SIR
(b)
Simulation Results
In this Section, simulation results obtained in different scenarios are presented. The
algorithm previously described has been tested in several realistic multi-target scenarios.

The proposed road environment is again of the type shown in Fig.11, i.e. a four-lane two
ways road, each lane being 4 m large. The radar under test, and any other interfering
radar in the explored region, scans an area 150 m long and 30 wide, by 15 steps of 2
each. The availability of radars with very narrow radiation diagrams reduces the risk of
blinding: when the signal from the interfering radar is detected along a direction different
from that of maximum gain, the interfering effect is reduced. The more the misalignment
between the two radar beams, the less the impact on the correct detection capability.
Considering also that the relative positions of the vehicles proceeding in opposite
directions change continuously, and rapidly, the blinding effect, whose duration is limited
in time, can be compensated by on board processing. However, investigation of the most
critical operation conditions remains a valuable task: to verify that the radar has a good
response even when the detected signal is strongly degraded, gives some guarantee that
the system will be able to react in time, by correlating the information collected over
multiple scans, mostly in terms of differences emerged.
A pictorial way to represent the result of a radar scan is to use 2D or 3D maps that, by
plotting the correlation values as a function of the spatial coordinates, permit to localize
directly the position of the detected vehicles. It is important to stress again that the
correlation peaks are extracted one by one, according to the algorithm presented above.
For this reason, as will be shown afterwards, they appear as normalized graphs, where all
the peaks identifying vehicles have almost the same altitude. At the same time, it is also
important to remind the fundamental role of the threshold optimization, as false peaks can
artificially result from the algorithm if the threshold value is not correctly set.
First set of Simulation Results: Ideal Antenna
In the first implementation it is reasonable to think that only a small percentage of cars
will be equipped with automotive radars; for this reason, in our first simulated scenario
(see Fig.13), we have the presence of only a few systems of this type; Fig.13, for
example, is quite similar to Fig.11, but there are two interfering radars.
Fig.13. Example with two interfering radars.
In the first simulation, Gold sequences with length NG = 31, 63 and 127 respectively, and
Lorenz chaotic sequences with length NL = 124, 252, 508 and 1020, are tested.

First of all, coherently with the analysis above, it is important to remark the importance
of threshold optimization.
Fig. reports a 2D map obtained by using chaotic sequences with N L = 124, for the
scenario of Fig.13, and assuming a threshold 2 times larger than the correlation standard
deviation. In this 2D map (and more generally in all the maps shown below), the color
intensity is related to the correlation peaks amplitude: the red spots correspond to higher
peaks, whereas the blue color corresponds to almost null correlation values. Moreover, a
mask, representing the vehicular scenario, is overlaid to the map, to immediately locate
the vehicles detected and those remaining undetected. Because of the small threshold, an
impressively high number of spurious spots (i.e. false detections) appear in Fig.14, that
make the map quite useless.
Fig.14. 2D map for the scenario in Fig.13, using chaotic sequences with NL = 124 and not
optimized threshold.
Fig.14 shows the 3D map corresponding to the simulation in Fig.13. The correlation
peaks should correspond to the vehicles that have been detected. But in this case the high
number of visible peaks is due to the interfering radars, and the threshold that is too low;
therefore, these peaks correspond to false detections, and this enforces the importance of
selecting an optimized threshold value.
Fig.15Error! Reference source not found.. 3D map for the scenario in Fig.13, using
chaotic sequences with NL = 124 and not optimized threshold.
By choosing the values reported in Table 7, we can minimize the number of false
detections, as shown in Figs. 16-21.

Fig.16. 3D map for the scenario in Fig.13, using chaotic sequences with NL = 124 and the
optimized threshold.
Fig.17. 3D map for the scenario in Fig.13Fig., using Gold sequences with NG = 31.
Fig.18. 3D map for the scenario in Fig.13, using chaotic sequences with NL = 252.
Fig.19. 3D map for the scenario in Fig.13 using Gold sequences with NG = 63.

Fig.21. 3D map for the scenario in Fig.13, using Gold sequences with NG = 127.
The set of simulation results discussed above refers to the almost ideal case of radar
antennas with negligible side lobes. This implies that, for each interfering radar, the
interference is maximum at the scan position (1 out of 15) for which the radar under test
is directed towards the interfering one, and null outside. Reciprocally, the worst condition
happens when the interfering signal emits along the direction of maximum gain for the
radar under test. Actually, an upper bound for the probability this event occurs can be
roughly estimated as (1/15)2. Figures 16-21 evidence the existence of a blinding zone
(clearly indicated in Fig.22), caused by the interfering radar; in practice the system
recognizes, at the right distance, 7 out of 10 vehicles, while the cars equipped with the
interfering radars, and the one behind the second of them, are not detected. In this
scenario, all the tested sequences actually detect the same seven vehicles, they cannot
detect the vehicles equipped with the interfering radar and the vehicle behind them.
Therefore, the performance of the two kinds of sequences is almost equivalent in this
idealized scenario.

Second Simulation Set: Antenna with Side Lobes
Lets now remove the hypothesis on the antenna, and assume that the interference signal
is also captured, though attenuated, along directions different from the one corresponding
to maximum gain (because of the antenna side lobes). In this case, it is possible to verify
that the chaotic sequences outperform the Gold sequences, at a parity of the bandwidth
expansion factor, which is coherent with the previous analysis.
Let us consider again the road environment in Fig.11, that includes only one interfering
radar; let us suppose that the radars are aligned, but the interfering signal is not zero also
in directions different from that of maximum gain. In details, the interfering signal is:
- attenuated by 10 dB in the direction 2 away from the maximum,
- attenuated by 20 dB in the direction 4 away from the maximum,
- attenuated by 30 dB in the directions that differ by more than 4 (in modulus)
from the maximum.
This situation, that however still represents a case of blinding, is less favorable than the
previous one (even though the number of interfering radars has been reduced), because of
the increased interfering signal. The 3D map of the correlation function is shown in
Fig.23: in this simulation, Lorenz chaotic sequences with (almost) the same 90% power
bandwidth occupancy of Gold sequences with NG = 31 are considered, i.e. a less
conservative (than before) bandwidth definition; under this assumption, N L = 200 must be
set for a fair comparison.
Fig.23. 3D map for the scenario in Fig.11, using chaotic sequences.
There are 5 vehicles clearly detected through high correlation peaks, while the lower peak
in the centre of the map corresponds to the vehicle at distance 90 meters along lane 2;
because of the value set for the threshold, that is necessary to avoid false detections, this
peak is difficult to identify. The 3D map obtained when Gold sequences are applied is
significantly worse; this is shown in Fig.24. Only 4 vehicles are detected, the blind zone
extension is much wider and, even more evident, the map is more contaminated than
before, so that distinguishing the peaks is more difficult.

Fig.24. 3D map for the scenario in Fig.11, using Gold sequences with NG = 31.
Third Simulation Set
Let us consider the road environment depicted in

Fig.25, where a single interfering radar is present, which is very close to the useful radar;
let us suppose that the radars are aligned, but the interfering signal is not zero also in
directions different from that of maximum gain, in the same conditions assumed before.
Fig.25. Scenario for the third simulation.
This situation, again dealing with a blinding event, is less favorable than the previous
one, because of the reduced distance of the interfering radar. In this simulation, only
chaotic codes are evaluated, in particular their ability to fight the interfering noise by
increasing the sequence length, and therefore the bandwidth. Figures 26-28 represent the
3D maps obtained by using Lorenz sequences of length NL = 252, 508 and 1020,
respectively.
Fig.26. 3D map for the scenario in

Fig., using chaotic sequences with NL = 252.

Fig.25, using chaotic sequences with NL = 508.

Fig.25, using chaotic sequences with NL = 1020.
The maps show that, by increasing the bandwidth, it is possible to remarkably increase
the detection ability of the radar; in fact the number of vehicles detected passes from 3 to
6, by increasing the sequence length from NL = 252 to 1020.
Finally, in Fig.29 the same scenario, but without interfering radars, is considered: in this
favorable situation all peaks are correctly detected (the one referred to the farthest
vehicle, which is the last one to be detected through the identification iterative algorithm,
is smaller). This is the map we can obtain, even in the presence of interfering radars,
during the time intervals when the interfering signal is weak, because of the misalignment
of the antenna beams.

It is expected this condition occurs for most of the time, thus preserving, on average, the
correct behavior. The worst conditions discussed before result critical to be overcome
through the combination algorithms applied by the on board processor. Reducing the
impact of the criticalities, as permitted by the proposed set of chaotic spreading
sequences, simplifies the computational complexity, and eventually makes processing
more feasible and reliable.
PRELIMINARY RESULTS ABOUT DE BRUIJN SEQUENCES FOR DS-SS

RADARS
To estimate the applicability of De Bruijn sequences in the DS-SS radar context, we start
from the evaluation of their auto-correlation properties, and the comparison with M- and
Gold sequences of similar length. Given the complexity of the generation process for the
De Bruijn sequences, we limit the discussion to a subset of De Bruijn sequences of period
N = 32 (n = 5). This choice allows a fair comparison with M-sequences and Gold
sequences of length 31.
The auto-correlation profiles of the M-sequences and Gold sequences are well known. A
De Bruijn sequence exhibits a satisfactory behavior, especially if compared to a Gold
(NG=31) one: for k 0, the normalized autocorrelation values are, on average, smaller
than those assumed by the Gold sequence, as shown in Fig.30.
Sequences with good auto-correlation properties are useful in several contexts; the
goodness of a sequence may be expressed through properly defined metrics such as the
Root-Mean-Square (RMS) sidelobe level of the auto-correlation function of a sequence a,
defined as:
1/ 2
1 N 1 2
P (a)
A, RMS
PA (i ) ( a ) (18)
N 1 i 1
Fig.30. Normalized auto-correlation curves for three sample sequences: Gold and M-
sequence of length 31, and a De Bruijn sequence of length 32.
The results obtained for the sequences of interest are summarized in Table 8. The M-
sequences provide the best behavior in terms of metric minimization, however, they also

provide the smallest set of available sequences. It is worth noting that the whole family of
2048 De Bruijn sequences of span n = 5 provides a lower expected value of P ( a ) A, RMS
than the family of 33 different Gold NG=31 sequences. It is possible to select, by
numerical verification, a subset of the De Bruijn sequences of span 5 which provide
closer agreement to random sequences, with respect to M- and Gold sequences, in terms
of P ( a ) A, RMS . Cross-correlation properties of the sequences were also studied. The De
Bruijn sequences show the highest maximum absolute value of the cross-correlation, a
zero average value, and a standard deviation that is very close to the one provided by the
other sets of sequences. The maximum absolute value of the cross-correlation for the De
Bruijn sequences, which equals the peak auto-correlation value, is due to the coexistence,
in a complete family of span n sequences, of different sequences, and their corresponding
complementary ones. If the complementary sequences are excluded from the set, the
maximum absolute cross-correlation value is reduced, without affecting the cardinality of
the set, which remains huge. In the vehicular radar application context, this consideration
would motivate the selection of some De Bruijn sequences, within a whole set, to use as
radar signatures, in order to lower the maximum absolute cross-correlation as much as
possible, and reduce the probability of false detection.
Table 8. Expected value of the RMS sidelobe level of the auto-correlation function, for
the test sequences.
(a)
Avg. P A, RMS
M-sequences 1
Gold NG=31 5.34
De Bruijn 4.67
CONCLUSION
Chaotic signals are of potential interest for application in long-range automotive radars
based on DS-SS. Their performance is often better than that of Gold codes with an (almost)
identical bandwidth expansion factor. On the other hand, they exhibit the important feature,
for these kinds of applications, of maintaining favorable correlation properties over a large
set of spreading sequences. These conclusions have been drawn by considering typical,
although necessarily simplified, road scenarios. The major problem in these kinds of
systems concerns the blinding effect due to an interfering radar proceeding in the opposite
direction; SS systems offer better performance from this point of view than that ensured by
more classical schemes. The chaotic solution does not seem much more complicated to
implement, and it is even more robust against interference.
In fixing the performance, an important role is played by the correct positioning of the
threshold against which the correlation function is compared. We have suggested a rule of
thumb, which is easy to implement and yielded very good performance in the considered
situations. However, the design of effective algorithms that are able to optimize the
threshold value in any operation condition remains a valuable open issue.

The behavior exhibited by the set of De Bruijn sequences tested, in terms of auto- and cross-
correlation properties, is encouraging, and motivates the efforts aimed at generating the
complete set of De Bruijn sequences of span n, for even greater values of n, and performing
an exhaustive evaluation of their properties.
REFERENCES
Andrenacci S., Gambi E., Sacchi C., Spinsante S., Application of de Bruijn sequences in
automotive radar systems: Preliminary evaluations, Proc. of 2010 IEEE Radar
Conference, 2010, pp. 959 964.
ASV project, Advanced Safety Vehicles in Japan, information available at

http://www.mlit.go.jp/jidosha/anzen/asv/index e.html, Mar. 2007.
Biondi F., Chiaraluce F., Gambi E., On the performance of chaotic sequences in
multiple access systems with variable spreading factor, Proc. ISITA 2004, Int. Symp.
Inform. Theory Appl., Parma, Italy, pp. 388-392, Oct. 10-13, 2004.
Callegari S., Rovatti R., Setti G., Embeddable ADC-based true random number
generator for cryptographic applications exploiting nonlinear signal processing and
chaos, IEEE Trans. Signal Processing, vol. 53, no. 2, pp. 793-805, Feb. 2005.
Chiaraluce F., Gambi E., Garello R., Pierleoni P., Performance of DCSK in multipath
environments: a comparison with systems using Gold sequences, IEICE Trans.
Fundamentals, vol. E85-A, no. 10, pp. 2354-2363, Oct. 2002.
ECC Report 23 Compatibility of Automotive Collision Warning Short Range Radar

Operating at 24 GHz with FS, EESS AND Radio Astronomy, Cavtat, Croatia, May
2003, available at http://www.craf.eu/ecc.htm.
ETSI EN 301 091-1 V1.2.1, Electromagnetic Compatibility and Radio Spectrum

Matters (ERM); Short Range Devices; Road Transport and Traffic Telematics (RTTT);
Radar Equipment Operating in the 76 GHz to 77 GHz Range; Part 1: Technical
Characteristics and Test Methods for Radar Equipment Operating in the 76 GHz to 77
GHz Range, Nov. 2004.
ETSI EN 302 288-1 V1.2.1, Electromagnetic Compatibility and Radio Spectrum

Matters (ERM); Short Range Devices; Road Transport and Traffic Telematics (RTTT);
Short Range Radar Equipment Operating in the 24 GHz Range, Dec. 2005.
ETSI TR 102 263 V1.1.2, Electromagnetic Compatibility and Radio Spectrum Matters
(ERM); Road Transport and Traffic Telematics (RTTT); Radio Equipment to be used in
the 77 GHz to 81 GHz Band; System Reference Document for Automotive Collision
Warning Short Range Radar, Feb. 2004

FCC Rules Part 15.253, Operation within the Bands 46.7-46.9 GHz and 76.0-77.0
GHz, 2 Jan. 2006.
Gambi E., Chiaraluce F., Spinsante S., Chaos-Based Radars for Automotive
Applications: Theoretical Issues and Numerical Simulation, IEEE Trans. Veh. Tech.,
vol. 57, no. 6, pp. 38583863, Nov. 2008.
Jansen C.J.A., Franx W.G., Boekee D.E., An efficient algorithm for the generation of
DeBruijn cycles, IEEE Tran. Inf. Th., Volume: 37, Issue: 5, Publication Year: 1991,
Page(s): 1475 - 1478.
Lon D., Balkir S., Hoffman M.W., Prez L.C., Robust chaotic PN sequence generation
techniques, Proc. IEEE ISCAS 2001, Sydney, Australia, vol. 1, pp. 53-56, May 2004.
Mayhew, G.L., Auto-correlation properties of modified De Bruijn sequences, Proc. of

IEEE Position Location and Navigation Symposium, 2000, pp. 349-354.
Mayhew, G.L, Third Update of the Order 7 De Bruijn Weight Class Distribution, Proc.
of 2007 IEEE Aerospace Conf., Big Sky (MT), March 3-10, 2007, pp. 16.
Morantes D.S., Rodriguez D.M., Chaotic sequences for multiple access, IEEE
Electron. Lett., vol. 34, no. 3, pp. 235-237, Feb. 1998.
Proakis, J.G., Digital Communications, McGraw-Hill, 1995
Rovatti, R., Mazzini G., Setti G., On the ultimate limits of chaos-based asynchronous
DS-CDMAI: basic definitions and results, IEEE Trans. Circuits and Systems I, vol. 51,
no. 7, pp. 1336-1347, July 2004.
Rovatti, R., Mazzini G., Setti G., Performance of chaos-based asynchronous DS-CDMA
with different pulse shapes, IEEE Commun. Lett., vol. 8, no. 7, pp. 416-418, July 2004.
SAFETEA-LU, Safe, Accountable, Flexible, Efficient Transportation Equity Act: A

Legacy for Users, information available at www.fhwa.dot.gov/safetealu/index.htm, Mar.
2007.
Sivanesan, K., Beaulieu, N.C., Performance analysis of band limited DS-CDMA

systems with diversity receivers over Nakagami fading channels, Proc. IEEE MILCOM
2004, Monterey, CA, pp. 1206-1212, Oct. 31-Nov. 3, 2004.
Sobhy, M.I., Shehata, A.R., Chaotic Radar Systems, IEEE MTT-S Digest, 2001, pp.
1701-1704.
Spinsante S., Andrenacci S., Gambi E., Modified de Bruijn Sequences for Spread

Spectrum Communications Proc. of 2011 IEEE Aerospace Conference, Big Sky, USA.
Wenger, J., Automotive MM-Wave Radar: Status and Trends in System Design and
Technology, Automotive radar and navigation technique (ref. No. 1998/230), IEE
Colloquium on 9 Feb 1998, pp. 1/1-1/7.

Monitoring and Controlling Product and Service Engineering:
The role of the Vehicles Repair Solutions
Nouha Taifi
ABSTRACT
This chapter presents and discusses the role of vehicles repair solutions in monitoring and
controlling products and services engineering. While vehicle repair solutions are needed
for the optimization of the vehicles of the customers, the products and services
engineering lead to the creation of the vehicles and the repair solutions. The chapter
argues that through vehicle repair solutions execution, the product and service
engineering can be controlled and monitored. That is, new knowledge can be shared for
the engineering of new products and services. The author objectives through this chapter
are to provide proofs of the importance of vehicles repair solutions in the vehicles
products and services engineering. Thus, she presents the vehicle repair solutions process
of a large network of after-sales services organizations in the automotive industry and the
resources of these contributing to the control and monitoring of the products and services
engineering of a large automotive company. The results of the study are based on an
intensive research and analysis in which the author interacted and interviewed various top
and middle managers in both the large network of after-sales services organizations and
the automotive company. The theoretical and managerial implications of this paper are
highly significant for the data and information about the vehicles repair solutions and the
product and service engineering are showing the role of vehicles repair solutions in the
monitoring and controlling of the products and services engineering in the automotive
industry.
Keywords: Product and service engineering, Innovation and quality

management,
INTRODUCTION
In the automotive industry, firms are urged to acquire new information and knowledge
from internal or external environment in order to sustain competitive advantage and to
pace with the increasing complexity of the environment. Their internal or external

capacities rely on various actors as research and development, suppliers, marketing,
manufacturing, customers and others (Afuah, 2003) and the product and services
development process consists in various phases going from design, testing,
manufacturing and after-sales services managing the lifecycle of these products and
services. The information and knowledge related to the engineering aspects of the
products and services are managed through this lifecycle management process (Amann,
2002).
The various phases of the product development process in the automotive industry (Clark
and Fujimoto, 1994) are occurring one after the other or managed in a concurrent manner
(Rouibah and Caskey, 2000) for the objective of decreasing time to market, customer
satisfaction achievement and high economic performance. Also, these phases are inter-
connected through integrated information and knowledge management systems. The
product development strategy related to concurrent engineering and integrated systems
represents one of the success factors of the firms in the automotive industry. And, the
modeling of the product development process and the evaluations of the different models
is essential for the continuous development of these processes (Smith and Morrow, 1999;
Huang and Gu, 2006; Formoso, 2002).
One of the other strategies of success of firms in the automotive industry is the new
product development process. That is, the creation of new knowledge, new products and
new services through innovation management. For instance, according to Haque and
Pawar (2002), the use of process modeling and analysis is primordial for the
improvement of the concurrent new product development process management. Also, the
concurrent engineering and integrated systems performance in the product lifecycle
process can play a major role in the identification of key players for the new product
development process.
The aim of this book chapter is to investigate on the role of one of the players in the
product lifecycle management, which is the after-sales services -more precisely the
vehicle repair solutions- in the new product development process. The objective of the
chapter is to identify the vehicle repair solutions factors and characteristics contributing
to the development and engineering of new products, new services and new processes
and the methodology used for that. That is, through the identification of the contributions
and the methodology, the development and engineering can be improved, controlled and
monitored for to end up with new products, new processes and new services.
The book chapter is organized as follows; the next section describes and explains the
process of product development to have a clear idea about its methodologies, systems and
technologies and so to use these explanations as a support for the investigation of this
research. The third section presents the research framework which consists in the
explanation of the innovation management in general and the new product development
process with a focus on the engineering aspects of the products, services and systems.
Also, the research framework identifies the purpose of innovation and new product
development through the elaboration on the goals of optimization, performance and

improvement in the organizational environments. The fourth explains the research
method which is a case study in the automotive industry about the after-sales services and
the last sections presents the findings, discussions and a model for controlling and
monitoring the products and services engineering.
PRODUCT AND SERVICE DEVELOPMENT PROCESS
According to Clark and Fujimoto (2000), the product development process consists in
various phases that go from design to maintenance. These phases are also named the
product lifecycle management (Saaksvuori and Immonen, 2002). These phases consist
also in various types of actors responsible of the different activities in these processes,
and of systems and technologies. According to Fairlie-Clarke and Muller (2003), in order
to ensure a successful concurrent engineering and resource allocation, a representation of
the product development process activities is necessary. Also, Morgan and Liker (2006)
have studied the Toyota product development systems and compared it with another
automotive company so to end up with the methodologies used for the product
development in the knowledge management, technology integration and multiple
functions collaborations fields.
And, Ahmadi et al. (2001) and Jun and Suh (2008) propose operational and strategic
frameworks for structuring product development processes to reduce time to market and
costs of production, and Sakao and Lindahl (2010) present various examples showing the
product/service-system design of the product development process; The systems,
technologies and processes used in the product development process determines the
effectiveness of the product development process and according to Holman et al. (2003),
a better information management focus leads to a higher product development
performance.
Fig.1. The concurrent role of the product development phases
Design and
Testing
Manufacturing and
After-Sales Services

Design and Testing
The design and testing phases (Figure 1) of the product development process are one of
the most crucial phases of the development of the products; it is used for the analysis of
the assembly feasibility that is essential for the manufacturing phase (Corallo et al.
2010a). It is a system as a business model involving various types of actors as designers,
engineers. At this stage, the product or service, before manufacturing, is designed through
computer-aided design tools and systems and also tested before final manufacturing;
There are also technologies used for to create prototypes and a collaborative
environments for the optimization of the design of the assembly of the products (Nan et
al. 2008).
Although these phases are the first ones of the product development process, through
concurrent engineering (Kusiak, 1993; Corallo et al., 2010b), they can become a part of
middle phases of the process of development for in case various changes are made to the
operations of the product development or components design (Alting, 1992); This latter is
related to the design checking and further development of prototypes of the products after
approval of the design. The engineering and testing of the products and services have
specific characteristics in which creative methods and systems of thinking as Design for
X (Pahl, Beitz and Wallace1996) lead to the ideas generations and problem-solving for
the design and testing.
Manufacturing and After-Sales Services
The manufacturing phase (Figure 2) of the product development process is the phase in
which there is a mass production based on the information and knowledge acquired from
design, testing and prototyping and from the computer-aided manufacturing tools and
systems used for manufacturing; . According to Tan et al. (2007), the value created in the
manufacturing phase of the product development process is now not only the production
of goods but also the creation of value through the identification of the utility of products
manufactured; The manufacturing companies focuses on the provisions of products and
their additional services (Haiyan and Yefeng, 2009).
Moreover, as a part of the product development process, the after-sales services are
provided to the customers as vehicle repairs and are based on the use of systems and tools
for diagnosis and problem-solving of the problems in the vehicles. These after-sales
services are a part of the product lifecycle management and benefits from the creation of
new managerial methodologies for continuous value creation for the after-sales services
and the manufacturing organizations (Taifi et al., 2010); The concurrent engineering and
integration of the entire product development process phases and steps is leading to faster
and easier products and services launch in the market and higher economic performance
and customer satisfaction.

RESEARCH FRAMEWORK
Fig. 2. Research framework
Quality
Management Products
Services Improvement
Systems Development
Processes
Innovation
Structures
Management
Quality Management-based Optimization and Improvement
The management of quality in all aspects of the organizations is necessary. For that, there
are principles and procedures to follow for to keep an acceptable quality within the
environment of the products and services (Heizer, 1997). According to Silvestro (1997),
there is a debate on the differences and similarities among quality management for the
manufacturing and services; This author concludes that the manufacturing quality
management learns from the service quality management as this latter is better
understood in management sciences, however, Gowen and Tallon (1999) states the
opposite. That is, the quality management systems of manufacturing are better developed
than service companies.
The main purpose of quality management is also to optimize and improve any process,
system, product or service in the environment for to continuously acquire high
performance; Quality management is used as a tool in the business process re-engineering
for continuous improvement of the processes (Muthu et al., 1999) and according to James
(1995), every enterprise needs to integrate quality management in its newly established
processes. Also, in business process re-engineering in which there are the move from an
as-is process from a to-be process, the main focus is exactly optimization and
improvement for to change radically (Hammer and Champy, 1993) to a better state of the
processes and continuously reach high competitive advantage.
Innovation Management as the Focus of Engineering
Innovation is the act of doing something for the first time and making it usable and
commercialized in a market. Various actors are involved in the innovation process as the
suppliers, customers, manufacturers, after-sales services and others (Afuah, 2003; Taifi
and Passiante, 2011) and the management of this innovation is based on the use of
information and communication technologies not only for the management of innovation
in the value chain but also in the supply chain; There are technologies supporting the act

of innovation and others supporting its transfer through the new product development
process and product development process. Also, there are technologies supporting and
improving concurrent new product development process among R&D laboratories
(Kowanga and Rasli, 2011).
Innovation and its management lead to strategic changes in the organizations (Tushman
and Anderson, 2004). For instance, there can be changes in the systems or processes, or
in the products and services and these changes are the results of engineering and re-
engineering through specific methodologies; Business process re-engineering is used in
the change of the processes of business (Muthu et al., 1999; Furey, 1993) and it can use
the matrix of change or the MIT process handbook. Besides, as stated by Wind and
Mahajan (1988), there is a need to review and re-engineer the new product development
process and to investigate in new methods of new product development.
RESEARCH OBJECTIVES AND METHOD
Based on the research framework regarding innovation management and quality

management, and the literature review on the product development process, the research
objectives are to demonstrate the important role of the after-sales services in the
automotive industry and more precisely the vehicle repair solutions in the process of
innovation and re-engineering. The specific objectives of the research are as follows:
- To define the actions undertaken by the repair phase in the new product development
process;
- To define the competences linked to these actions and the development method of these
competences;
- To define the dynamisms of the repair solutions actions in the new product
development process.
A single case study (Yin, 2003; Creswell, 2003) in the automotive industry was adopted
although this is rich of various types of organizations ranging from after-sales services to
design and manufacturing organizations. The case study data were collected through
interviews carried out over a period of six months in a face to face manner in the design -
manufacturing company, and a sample from a large population of after-sales services was
selected for the administration of two types of questionnaires. The interviewees have
various middle and top-management positions related to the quality of the services,
product development, manufacturing, and product and service engineering.
The interview objective was to identify the different types of relationships with the after-
sales services and the specific roles of the interviewees. The questionnaires were two-
fold; one for the identification and definition of the vehicle repairs solutions and the
methods used for the development of the competences linked to vehicle repair solutions,
and the other was for the identification of the method and characteristics of the
contributions of the vehicle repair solutions in the product and service engineering, and

was dedicated to a sample from the population of a large network of after-sales services
organizations.
FINDINGS
Actions of VRS in the NPSD process
The after-sales services small and medium enterprises provide vehicle repair solutions to
the customers of the design-manufacturing firm. Through a project that has the objective
of identifying and using the after-sales services firms expertise in vehicle repair solutions
some of them were invited to participate to a strategic community in order to explicit
their know-how in products, services, systems and processes in the automotive industry
(Taifi and Passiante, 2010). Through their collaboration in this community, they provide
information and knowledge on the level of quality of the vehicle repair solutions, on the
vehicles themselves and on new vehicle repair solutions (Taifi and Passiante, 2012).
The vehicle repair solutions are composed of a diagnosis tool for the identification of the
problems in the vehicles and a problem-solving system for repair. Thus, the after-sales
services firms provide information and knowledge on the quality of the diagnosis tool
and problem-solving system and on new types of diagnosis tools and problem-solving
systems. That is, they communicate the necessary information for the re-engineering of
the repair solutions and the improvement of their quality and the engineering of new
services. Also, they share information and knowledge on the quality of the vehicles that is
new information and knowledge for the engineering of new vehicles.
Competences Development for NPSD
The skills of the after-sales services firms are subject to developments through
professional trainings. These are courses, seminars and experiments regarding the vehicle
repair solutions and the vehicles themselves (Taifi and Passiante, 2011). During the
vehicle repair phases, the competences used are related to the after-sales services and are
technical with expertise in mechanical, electrical and electronic engineering. Thus, these
mechanical, electrical and electronic services competences are subject to improvement
and change when there are new vehicles on the market through the provision of trainings
in various expertise linked to the vehicles and the repair solutions. Also for the vehicles,
the skills of the after-sales services firms are developed not only on each type of
component but also on the material used for manufacturing.
The professional training for product and service optimization in the case of vehicle
repair solutions is dedicated to this learning organizations that are continuously in
learning processes to reach high performances (Senge, 1990). Also, through the
professional training, as a strategy for organizational learning (Argyris and Schon, 1978),
the expertise improvement on the vehicles and vehicles repair solutions of the strategic
community of after-sales services, contribute to innovation and new product

development. That is, as there are new professional trainings as there are more
competences and skills for product and service engineering.
CONCLUSIONS: A MODEL AND MANAGERIAL IMPLICATIONS
Controlling and Monitoring Product and Service Engineering: A model
The product and service engineering in the design and manufacturing phase of the
product development implies the participation of various actors and in this model the new
potential innovator is the vehicle repair solutions. This latter is rich of information and
knowledge on different components of the products and on the services for the products
(Figure 3). Through the new product development process, the vehicle repair solutions as
a part of the after-sales services are integrated in the monitoring and control of the
product and service engineering. For that, in order to strengthen and to enhance
collaboration, organizational structures as strategic as the objective of their creation are
necessary and strategies for competences development on the products and services as
well.
Fig. 3.: Controlling and Monitoring Product and Service Engineering
System
s
Service
s
Product
s
Vehicle
repair
solutions
Controlling and Monitoring
The engineering of products, services and systems based on the use of vehicle repair
solutions depends thus on two main dynamisms that are the organizational structures for
product and service engineering and the professional training. That is, as there are new
competences development programs as there are the possibility for the vehicle repair
solutions to contribute to product and service engineering and as there are new
monitoring and control from the part of the vehicle repair solutions as there are new
products and service engineering (Figure 4). Also, a part from being high-technological-
based, the product and services are also high technological.

Fig. 4.: The role of professional training in product and service engineering
Increasing Increasing
number number
of new products of new products
and services and services
Increasing number of Increase of monitoring and

the professional control of product and
training service engineering
Managerial Implications for Product and Service Engineering
In order to achieve customer satisfaction and high economic performance, the firms and
organizations are continuously searching for new methods of collaboration and new
innovators to create new information, knowledge and thus new product, processes,
services and systems. The research in this paper led to the creation of a model for the
integration of the vehicle repair solutions in the new product development and more
precisely the control and monitoring of the product and service engineering.
Fig. 5: The concurrent role of professional training and organizational structures
Professional
training
New information,
knowlegde, products,
Organizational services Control and
structures monitor product
and service
engineering

From the managerial perspective, it also leads to the development of guidelines for the
continuous integration of repair solutions in the product and service engineering and thus,
for instance in the case of aerospace industry in which these guidelines of repair solutions
can be used to monitor aerospace products and services engineering or in the case of the
software industry in which repair solutions are used as well. As follows are the guidelines
that can be used by practitioners in different industries (Figure 5):
- As the awareness of the practitioners about the importance of repair solutions in product
and service engineering is set up, as organizational structures -as strategic communities-
for knowledge management can be created; then,
- As programs for competences development are organized, as these can be used for to
support the repair solutions in the development of new products and services; also,
- As the objective of products and service engineering is clearly in mind of repair

solutions, through the use of competences development and communities for
communication and discussions, as the role of repair solutions is more powerful in
monitoring and control of the product and services engineering.
REFERENCES
Afuah, A. (2003), Innovation Management: Strategies, implementation and profits, Oxford

university press.
Ahmadi, R., Roemer, T.A. and Wang, R.H. (2001), Theory and Methodology: Structuring
product development processes, European Journal of Operational Research, 130, pp. 539
558.
Alting, L. (1993), Lice-cycle design of products: A new opportunity for manufacturing

enterprises, In Kusiak, A. (Eds): Concurrent engineering: automation, tools and techniques,
Wiley: Interscience.
Amann, K. (2002), Product Lifecycle Management: Empowering the Future of Business, CIM
Data Inc..
Argyris, C. and D.Schon, 1978, Organisational Learning: A Theory of Action Perspective,

Reading, MA.
Clark, K.B. and Fujimoto, T. (1991), Product development performance, Strategy,

Organization, and Management in the World Automobile Industry, Boston, MA: Harvard
Business School Press.
Corallo, A., Magherita, A. and Pascali, G. (2010a), Digital mock-up to optimize the
assembly of a ship fuel system, Journal of Modelling and Simulation of Systems, Vol. 1,
Issue 1, pp. 4-12

Corallo, A., Lazoi, M., Taifi, N. and Passiante, G. (2010b), Integrated systems for
product design: the move toward outsourcing. In Nunes, M.B., Isaias, P. and Powell, P.
(eds): Proceedings of the IADIS international conference on information systems. pp.
216-222, 18-20 March, Porto, Portugal.
Creswell, J. W. (2003), Research Design: Qualitative, Quantitative, and Mixed Methods

Approaches, Sage publication.
Fairly-Clarke, T. and Muller, M. (2003), An activity model of the product development

process, Journal of Engineering Design, Vol. 14, N. 3, pp. 247-272.
Formoso, CT, Tzortzopoulos, P and Liedtke, R. (2002), A model for managing the product
development process in house building, Engineering Construction and Architectural
Management, Vol. 9, N. 5-6, pp. 419-432.
Furey, Timothy. R., 1993, A Six Step Guide to Process Reengineering, Planning Review,
21(2), pp. 20-23.
Gowen, C.R. and Tallon, W. J. (1999), Quality management practices in Manufacturing and
Service Corporations: How are they different?, American Journal of Business, Vol. 14, N. 1,
pp. 33-40.
Haque, B. and Pawar, K.S. (2001), Improving the management of concurrent new product
development using process modeling and analysis, R&D Management, Vol. 31, Issue 1, pp.
27-40.
Haiyan, Z. and Wang, Y., 2009, The Product - Service Bundling Model Within Manufacturing
Sectors: An Empirical Study From China, Proceedings of International Conference on
Management Science and Engineering.
Hammer, M. and Champy.J. (1993), Reengineering the Corporation: A Manifesto for Business
Revolution., Harper Collins, London.
Heizer, J.H. (1997), Cases in Total Quality Management: Manufacturing and Services, South-
Western Educational Publishing.
Holman, R., Kaas, H.W. and Keeling, D. (2003), The future of product development,
McKinsey Quarterly, N. 3., pp. 28-39.
James,M. (1995), The Great Transition: Using the Seven Disciplines of Enterprise
Engineering to Align People, Technology, and Strategy., American Management Association,
New York.

Jun, H.B. and Suh, H.W. (2008), A Modeling Framework for Product Development Process
Considering its Characteristics, IEE Transactions on Engineering Management, Vol. 55, Issue:
1, pp. 103-119.
Huang, H.Z. and Gu, Y.K. (2006), Modeling the product development process as a dynamic
system with feedback, Concurrent Engineering, Vol. 14, N. 4, pp. 283-291.
Kowang, T.O. and Rasli, A., 2011, New product development process in multi-location R&D,
organization: A concurrent engineering approach, African Journal of Business Management,
Vol. 5, (6), pp. 2264 2275.
Kusiak, A. (1993), Concurrent engineering: automation, tools and techniques, Wiley:

Interscience.
Muthu, S., Whitman, L. and Cheraghi, H. (1999) Business process re-engineering: A

consolidated methodology, Proceedings of the 4th Annual Conference on Industrial
Engineering Theory, Applications and Practice, November 17-20, San Antonio, Texas, USA.
Nan, F-Q, L, T-T; Zhang, Y-L; Wang, H-F; Zhang, L-L, 2008, Research on collaborative
technology of networked collaborative product digital pre-assembly system. Journal of
system simulation, 20 (23), 6418 6423.
Pahl, G., Beitz, W. and Wallace, K. (1996), Engineering Design: A systematic approach,
Springer.
Rouibah, K. and Caskey, K.R., 2000. Managing concurrent engineering with early
supplier involvement: a case study, In International Journal of Computer Integrated
Manufacturing, Vol. 18, N. 6, pp. 509-521.
Sakao, T. and Lindahl, M. (2010), Introduction to product/service-system Design. Springer.
Silvestro, R. (1997), The manufacturing TQM and service quality literatures: Synergistic or
Conflicting paradigms, International Journal of Quality and Reliability Management, Vol. 15,
N. 3, pp. 303 328.
Senge, P.M. (1990) The Fifth Discipline, Century Business, London.
Smith, R.P. and Morrow, J.A. (1999), Product development process modeling, Design Studies,
Vol. 20, Issue 3, pp. 237-261.
Taifi, N. and Passiante, G., 2011, The concurrent role of professional training and operations
management: Evidences from the after-sales services information systems architecture in the
automotive sector. In Chiaberge, M. (Ed), Trends and Developments in Automotive
Engineering, Chp 4, pp. 61-78. InTech publisher.

Taifi, N. and Passiante, G., 2012, Speeding up NPSD through strategic community
creation: Case of an automaker after-sales services partners, Service Industries
Journal,Vol. 32, N. 11, pp. 2115-2127, Taylor and Francis Group Publisher.
Tan, A.R., McAloone, T.C. and Gall, C. (2007), Product/Service system development An
explorative case study in a manufacturing company, International Conference on Engineering
Design.
Tushman ML, Anderson P. (2004). Managing Strategic Innovation and Change: A Collection
of Readings. Oxford University Press: NY
Yin RK. (2003). Case Study Research: Design and Methods; Thousand Oaks, CA: SAGE
Publications
Wind, Y. and Mahajan, V. (1988), New product development process: A perspective for
reexamination, Journal of Product Innovation Management, Vol. 5, pp. 304 310.

VHDL-AMS Simulation of an AHRS Platform for Automotive
Mariagrazia Graziano and Massimo Ruo Roch
INTRODUCTION
Entanglement between electronics and mechanics is a key aspect of future improvements

in automotive security systems (Casier et al. 1998). Attitude and Heading Reference
System (AHRS) platforms are an example. They are widely used in aeronautics (Gebre-
Egziabher et al. 1998), and currently adopted in Formula 1 races. Recently they are being
introduced in mass production vehicles.
They allow real-time detection of vehicle stability and trajectory with respect to a fixed
reference system. They are based on accelerometers, gyroscopes, magnetometers and
GPS receivers (Li et al. 2006). Accelerometers detect the acceleration along three main
axes. Gyroscopes are able to pick up the rotation around the three main axes. The
magnetometer measures the magnetic field along the three components. GPS detects the
instantaneous vehicle position. By means of a dedicated processor, this position is then
compared to the one estimated on the basis of data obtained through the other sensors (Li
et al., 2009). A combination of those data gives the final information required.
The AHRS platforms adopted in Formula 1 races are used to be aware at run time and
with very high precision of the position of the other cars in order to optimize the race
strategy. In this case data detected on board are sent through vehicle telemetry to team
software facilities. Data are then processed in combination with extremely detailed
information on the race circuit. Clearly, this type of organization is fruitful in Formula 1
as circuits detailed maps are easy to be generated (few, well known...). This is clearly
unfeasible for all the roads used by common vehicles. In this case GPS is used as an aid,
and, though not with the same precision obtained in Formula 1 races, an important help
can be given to on board security systems, in order to obtain a sort of automatic cruise
control.
The AHRS accuracy and timeliness are linchpins for a reliable automatic-cruise system
and, in particular, for a safety-guardian-on-board one. However, the AHRS sensors have
limited accuracy, and this might have an impact on an effective usage on board. It is
mandatory to understand how they are applied and used in case of a mass production
vehicle in several realistic conditions, and how errors due to sensors could impact
guidance systems performance.
Currently, several practical examples are found for avionics applications (Cordoba 2007,
Adiprawita1 2007). The use of a correct design methodology is becoming essential in
order to optimize AHRS performance for critical applications (Wen-shu et al. 2010).

Recently high performance electronic devices are being adopted in order to increase the
run-time processing capabilities (Zhang et al. 2007). However, in order to exploit the
potentialities of this system in a mass production vehicle, every aspect must be
considered and, if critical, balanced introducing the needed corrections at the design
time.
Instead of a trial and error methodology, thus, we propose to simulate the platform using
VHDL-AMS language. VHDL-AMS (1076.1, 1999) is a super-set of VHDL (VHSIC
Hardware Description Language), that, normally, is used to model digital circuits. The
AMS extension supports differential equations and algebraic constraints (Ashenden et al.
2002). Their values are assumed by electrical quantities that thus can describe analog
circuits as well as mixed-technology systems (Radezki, 2009). Mechanics and thus
electro-mechanical systems can be efficiently simulated (Graziano, Ruo Roch, 2008). The
essential aspect is that it is possible to capture the impact of mechanics on electronics
design choices (and vice-versa if necessary). Another interesting feature of this language
is multi-resolution (Mentor Graphics, 2004). Thanks to it, depending on the interest and
importance in the devices involved, different blocks of the system can be described using
different levels of abstraction. In the automotive field, for example, the car mechanical
system can be roughly described, while, on the contrary, electronic subsystems
description can be refined to the needed level of detail.
In this proposal we aim at exploiting VHDL-AMS multidisciplinary features in order to
evaluate how an inertial platform can be simulated and its possible critical feature
explored and assessed at design time. Starting from the work in (Barbero, 2006), we
model the AHRS system. We describe the vehicle behavior in order to obtain input
sources to the platform. These input conditions are related to real cases. i.e. the vehicle in
motion on a straight road at constant velocity, or on a bend at both small and high speed.
We show that the whole AHRS system could be well modeled using this method.
Moreover we show how this methodology allows to introduce errors in models and
analyze their effects on final performance. We then demonstrate how this methodology
allows achieving a good understanding of mechanics impact on electronics design
choices.
INERTIAL PLATFORM
One of the reasons why inertial platforms are more and more being successfully used is
the advance in sensor technology. Currently, most of the sensors are based on Micro
Electro-Mechanical Systems (MEMS). These systems are a mix of electronics
components based on integrated circuits and of mechanical micro-structures built thanks
to micro-machining. Compactness and precision are the main advantages of these
systems. A totally integrated MEMS based AHRS system could reach a size of 45mm^2.
In order to estimate a vehicle stability with a good precision, a few coordinates
transformation are necessary, starting from the single sensor reference system, up to the
terrestrial reference system. One of the most diffused method is using quaternions, which,
though not intuitive, allows a simpler transformation with respect to more traditional
Euler corners (Barbero, 2006). No details will be given here for what concerns this
transformations, as it is out of the focus of this chapter. However, a few hints on the

reference systems used will be given herein. In Fig. 1 a block diagram of the platform is
given. It is composed by three accelerometers, three gyroscopes, one magnetometer and a
processor with its memory. Every sensor has its own reference axis. These axes are
associated to the platform reference system (X board, Y board) . The platform is mounted
on a vehicle, so coordinates must be referred to the car body as well. Normally, the x axis
has longitudinal direction with respect to the vehicle and is oriented according to the
engine direction. The y axis is perpendicular to x on the plane of the ground, with
orientation in the right-hand outbound direction. The z axis is perpendicular to the ground
plane and is oriented toward ground (or oppositely, depending on the adopted
convention). This reference system is finally referred to the terrestrial reference system.
Fig.1. Inertial platform block diagram.
VHDL-AMS
The VHDL language models discrete systems, and is normally used for digital signals in
digital circuits. In case of analog and mixed-signal electrical design, or even mixed
electrical/not-electrical system, then VHDL-AMS should be the correct language used for
modeling. In fact it supports continuous signals derived by differential algebraic
equations, thanks to a dedicated analog solver. Several works report VHDL-AMS as an
effective simulation language for efficient design of complex mixed-signal electronics
systems (Christen & Bakalar, 1999). Many works can be found in the telecommunication
field. A Bluetooth transceiver model was described in (R. Ahola et al., 2003). A less
behavioral transceiver circuit was described using VHDL-AMS in (J. Oudinot et al.,
2003). A white Gaussian noise was added to the description of a RF block in (E. Normark
et al.).

An interesting top-down design methodology, validated by measurements, was proposed
in (V. Nguyen & Naviner, 2005) for the design of a delta-sigma modulator. First a coarse
description of components was given, after various error sources like jitter, thermal noise,
and capacitor mismatch were added to models. After a fabrication phase, measurements
confirmed both model accuracy and top-down methodology effectiveness. A
methodology for the design, the simulation and the fault analysis of a Controller Area
Network (CAN) bus is presented in (W. Prodanov & Buzas, 2009).
In (Santarini, 2006) VHDL-AMS is suggested for modeling complex electronic systems
interacting inside the automotive environment.
In (M. Gursoy & Pelz, 2008) a methodology is presented to investigate and predict the
effects of the interaction between CAN transceivers and twisted pair transmission line
connecting them in terms of electromagnetic emission.
A complete reference with several real cases is in (Radezki 2009).
The possibility to solve differential algebraic equations allows to model non-electronic
systems. In (F. Pecheux & A.Vachoux, 2005) several disciplines are involved in the
modeling of an airbag system. Electrical, thermal, optical, mechanical and chemical
knowledge are synthesized in the resulting model.
A further up-to-date example is in (Gao et al., 2010), where a multidomain dynamic
proton-exchange-membrane fuel-cell stack model is described and simulated using
VHDL-AMS.
A micro-electromechanical system piezoelectric micro-generator is described in (H.
Boussetta & Soudani, 2010) demonstrating how the language is particularly suitable to
accurately represent automotive electromechanical components.
It is worth noticing that the design of automotive systems requires a multitude of single
components and sub-blocks. At the same time, complete analyzes at system level are
needed. In (Moser & Mittwollen, 1998) the first results of a joint project between
automotive industry partners and tool makers are presented. The authors modeled a brake
system with ABS including hydraulics, mechanics and control aspects.
A fuzzy logic controller for active mixed mode suspension system modeling was
demonstrated in (Wang & Kazmierski, 2005).
As previously mentioned, the VHDL-AMS language has another key characteristic: it
allows to describe systems with different levels of abstraction. A top-down design
methodology is then suitable.
First, a preliminary behavioral description of components is used to obtain a coarse
functionality test of the system. Then, as progressive refinements define real circuit
performance, a detailed description is provided for the most important blocks of the
system.
This feature is particularly enhanced by ADMS (MentorGraphics, 2004). It allows to co-
simulate both high level VHDL-AMS architectures and Spice-level netlists using ELDO
in the same simulation environment. A few works demonstrated this method (M.R. Casu
et al., 2007) (M.R. Casu & Graziano, 2008).
In an automotive system, this multi-resolution capability implies that critical electronic
blocks can be roughly described at first. Later they are accurately refined, thus exploiting
the native VHDL-AMS strength in modeling digital and analog devices, if needed, down
to transistor level (Graziano and Ruo Roch, 2008).

Besides the obvious advantage from the point of view of the possible achievable results,
also a lightweight simulation from the point of view of CPU time can be obtained. It is
then of utmost importance the possibility to capture the effects of electronic details at the
system automotive level. Conversely, crucial is the capability of deriving electronic
device design constraints from car system required performance.
The electronic system will be thus effectively optimized, taking into account a real
mechanical environment. In this way reliability and fault test cases are reduced.
In this chapter, starting from the work in (Barbero, 2006), even if we do not go into
details of the electronic system, we show how it is possible to create, early in the
production phase of a AHRS platform, an assessment of constraints and faults effects
exploiting VHDL-AMS capabilities. It means co-simulating electronic and mechanical
devices/sensors to better describe their strictly related operations.
Another very important point consists in analyzing how the performance of such a system
is influenced by its real application environment.
MODELING
The model we implemented is a hierarchical one. The top level is the system organized in
three sub-blocks, as depicted in Fig. 2. The most important is the real AHRS platform
(SENSORS in the figure). This, in the real world receives inputs as accelerations and
magnetic field. In our model these quantities are inputs to the SENSORS, generated by a
PHYSICAL SOURCES block in which we model the needed electrical and mechanical
signals. The PHYSICAL SOURCES to the whole system are linear or angular
accelerations. They can be chosen among several types of signals at the beginning of a
simulation. Typical cases are: 1) a constant, 2) a sine wave, 3) a ramp, 4) pulse, 5) a step.
Other inputs are reference signals to the sensors and to the analog to digital converters.
The SENSORS outputs are digital, as they must be processed by a digital processor. In
order to appreciate in a unique quantity these output values, they are converted into
processed signals through the DATA-PROCESSING, unit, which transforms sensors
information in values that can be associated to the vehicle behavior.
Fig.2. System level of the whole platform model.

Our model does not include the real micro-controller and/or the GPS system. These,
though mandatory in a complete design, are not needed to demonstrate methodology
effectiveness.
The SENSORS block internal organization is depicted in Fig. 3. It is organized in three
identical sub-blocks, each for one of the axis.
Two kinds of inputs are sent to the block:
electrical signals, like reference voltages, power supply and a clock signal.
physical values like linear and angular accelerations.
For each block, two distinct outputs are available, for accelerometer and gyroscope
values, respectively. A magnetometer is not inserted in this version of the model.
Fig.3. Internal structure of the SENSORS block.
Each AHRS block, the core of the system, is internally organized as sketched in Fig. 4.
An accelerometer and a gyroscope, and their respective analog-to-digital converters
(ADC), generate 12 bits digital output signals. Inputs are the same electrical signals
described above, and the linear and angular acceleration values from the external system.
Fig.4. Internal structure of a single AHRS system.

It is now worth giving an example of the VHDL-AMS description of the AHRS basic
block. As in common VHDL, two are the main constructs used: entity and architecture.
The entity section follows.
entity ahrs is
generic(
constant ma : MASS := 0.0; constant ca : DAMPING := 0.0;
constant ka : STIFFNESS := 0.0; constant d0a : REAL := 0.0;
constant cap_par_a : REAL := 0.0; constant epsx : REAL := 0.0;
constant mg : MASS := 0.0; constant betag : DAMPING := 0.0;
constant rg : REAL := 0.0; constant d0g : ANGLE := 0.0;
nbit : INTEGER := 0; tD_conv : TIME := 0
);
port (
terminal motion_mass_accel : TRANSLATIONAL
terminal rot_mass_gyro : ROTATIONAL;
terminal alim_ag_up, alim_ag_down : ELECTRICAL;
terminal ref_ad_plus : ELECTRICAL;
terminal ref_ad_minus : ELECTRICAL;
signal ck : in std_logic;
signal dout_acc : out std_logic_vector(nbit-1 downto 0);
signal dout_gyro : out std_logic_vector(nbit-1 downto 0)
);
end ahrs;
This part gives the external view of the structure by assessing its ports, i.e., its inputs and
outputs. In simple VHDL these are usually signals, which are digital and assume
basically only values 0 and 1 (when valid) and are of type standard logic (std_logic).
Here clock (ck) and block outputs (dout_acc, dout_gyro) are signals, too. In VHDL-AMS
ports can also be terminals, that can assume continuous values of types that depend on
how they are declared.
In this case, for example, terminals used for power supply (alim_ag_up, alim_ag_down)
and for voltage references (ref_ad_plus, ref_ad_minus) are of electrical type.
The two physical inputs to the accelerometer (motion_mass_accel) and to the gyroscope
(rot_mass_gyro) are of translational (describe a linear movement) and rotational
(describe a rotation) type, respectively.
Just above the port description, a generic section is given, which describes parameters
used to configure internal elements. Their name, type and default values are specified.
When this ahrs block is instantiated in the SENSORS block, these parameters are
assigned, exactly like a value when a subroutine is called in a high level programming
language. The advantage of this organization is flexibility and generality. In fact, without
changing the hardware description, different parameters can be used for modeling
systems with same structure but different characteristics.
The second part of the VHDL-AMS description is the architecture which includes the
internal structure. In this case this block is composed of other sub-blocks. They must be
described in other specific files, and here they are first declared (see components
declaration section) as components and then instantiated and connected (see components
instance and connection section). The connection is made by using signals or terminals
previously declared (see internal signals or terminal declaration section). When a
component is instantiated, its parameters (generic's) are fixed, too. For example the
accelerometer has a mass parameter m. When instance acc of the accelerometer is

connected, the mass is associated to parameter ma. This is itself a parameter, declared as
a generic of the ahrs block. The value will be definitely fixed when this block will be in
turn used at a higher level of the hierarchy.
architecture behaviour of ahrs is

-- components declaration
component accelerometer
generic (
constant m : MASS := 0.0; constant c : DAMPING := 0.0;
constant k : STIFFNESS := 0.0; constant a : REAL := 0.0;
constant d0 : REAL := 0.0; constant cap_par : REAL := 0.0
);
port (
terminal motion_mass_accel : TRANSLATIONAL;
terminal e_up, e_mid, e_down : ELECTRICAL
);
end component;
component resistor
generic (r : REAL := 0.0);
port (terminal resp, resm : ELECTRICAL);
end component;
component gyroscope
generic (
constant m : MASS := 0.0; constant beta : DAMPING := 0.0;
constant r : REAL := 0.0; constant d0 : ANGLE := 0.0;
constant a : REAL := 0.0
);
port (
terminal rotation_mass_gyro : ROTATIONAL;
terminal e_g_up, e_g_mid, e_g_down : ELECTRICAL
);
end component;
component a2d_1
generic (
n : INTEGER := 0; Rref : REAL := 0.0; tD_out : TIME := 0 ns
);
port (
signal dout : out std_logic_vector(n-1 downto 0);
signal underflow : out std_logic;
signal overflow : out std_logic;
signal encode : in std_logic;
terminal ref_plus : electrical; terminal ref_minus : electrical;
terminal ain : electrical
);
end component;
-- internal signals or terminal declaration

terminal v_modul_acc : ELECTRICAL; -- modulated output of accelerometers
terminal v_modul_gyro : ELECTRICAL; -- modulated output of gyroscopes
signal o_flow_a, u_flow_a, o_flow_g, u_flow_g : std_logic;
-- components instance and connection

begin
acc : accelerometer
generic map( m => ma, c => ca, k => ka, a => epsx, d0 => d0a, cap_par => cap_par_a
)
port map( motion_mass_accel => motion_mass_accel, e_up => alim_ag_up,
e_mid => v_modul_acc, e_down => alim_ag_down);
r : resistor
generic map( r => 3.0e6 )
port map( resp => v_modul_acc, resm => electrical_ref );
omega : gyroscope
generic map( m => mg, beta => betag, r => rg, d0 => d0g, a => epsx )
port map( rot_mass_gyro => rot_mass_gyro, e_g_up => alim_ag_up,
e_g_down => alim_ag_down, e_g_mid => v_modul_gyro );
R2 : resistor
generic map( r => 3.0e6 )

port map( resp => v_modul_gyro, resm => electrical_ref );
a2d_acc : a2d_1
generic map( n => nbit, Rref => 10.0e6, tD_out => tD_conv )
port map(
ain => v_modul_acc, ref_plus => ref_ad_plus, ref_minus => ref_ad_minus,
encode => ck, overflow => o_flow_a, underflow => u_flow_a,
dout => dout_acc
);
a2d_gyro : a2d_1
generic map(n => nbit, Rref => 10.0e6, tD_out => tD_conv)
port map(
ain => v_modul_gyro, ref_plus => ref_ad_plus, ref_minus => ref_ad_minus,
encode => ck, overflow => o_flow_g, underflow => u_flow_g,
dout => dout_gyro
);
end architecture behaviour;
It is now interesting to go further into details of the accelerometer and gyroscope model.
The accelerometer (Ripka, 2001) can be modeled as a seismic mass connected through a
spring and a dashpot to the external case. This results in a one degree of freedom (1 DOF)
system. The mass is organized in two interdigitated structures like in Fig.5, where one is
fixed and the other is the mobile (seismic) mass. The motion of this mobile mass
generates a capacitance variation as depicted in Fig. 5 right detail. In absense of
acceleration along the detection direction, the capacitance across adjacent fingers is
constant and equal. If, on the contrary, an acceleration is imposed to the system, then the
capacitance between adjacent fingers will change, due to a displacement of the central
finger. C1 and C2 in Fig. 5, on the right, can than be used to build a capacitive divider, in
which the output to input ratio is a measure of the acceleration of the system.
Fig.5. Internal structure of the accelerometer mass. Right, a detail.
RESULTS
In Fig.6 a detail of the model is given. The sensor is fed with a differential signal, applied
to the upper (C1) and the lower (C1) finger capacitor. As a first approximation, we can
state that the voltage read at the output junction depends on the voltage given as input and
its combination between the two capacitances C1 and C2, according to the equation:

(C1 C2)(C1+C2)
Fig.6. Electrical model with a parasitic capacitance.
In case we want to take into account sensor non-idealities we can add then at this point of
the model. As an example, certainly a parasitic capacitance exists between the output
node and the reference. This can be accounted for by modifying the dividing ratio
according to the equation:
[(C1 C2)(C1+C2)][1(1 +Cp(C1+C2))]
The accelerometer VHDL-AMS description is in the following. After the entity

declaration the architecture is coded. It is organized in a first part where needed
unknowns for the equation description are declared, and in a second part where the
equations are the given. Each quantity is based on through and across variables that can
be assimilated to a current and a voltage in an electrical system.
entity accelerometer is
generic (
constant m : MASS := 0.0; constant c : DAMPING := 0.0;
constant k : STIFFNESS := 0.0; constant a : REAL := 0.0;
constant d0 : REAL := 0.0; constant cap_par : REAL := 0.0
);
port (
terminal motion_mass_accel : TRANSLATIONAL;
terminal e_up, e_mid, e_down : ELECTRICAL
);
end accelerometer;
architecture level0 of accelerometer is

-- variables declaration
quantity x across f through motion_mass_accel to translational_ref;
quantity v_up across i_up through e_up to e_mid;
quantity v_down across i_down through e_down to e_mid;
quantity v_par across i_par through e_mid to electrical_ref;
quantity q_up, q_down, q_par : CHARGE;
quantity d_up, d_down : DISPLACEMENT;
quantity cap1, cap2 : CAPACITANCE;

-- equations description
begin
-- differential equation for 1DOF description
x'Dot'Dot == (-k*x - c*x'Dot - f)/m;
-- Relative displacement between fingers
d_up == d0 + x;
d_down == d0 - x;
-- Capacitance variation due to displacement
cap1 == a*EPS0/(d0+x);
cap2 == a*EPS0/(d0-x);
-- Current generated by capacitance variation
q_up == cap1 * v_up;
q_down == cap2 * v_down;
i_up == q_up'Dot;
i_down == q_down'Dot;
-- parasitic terms
q_par == cap_par * v_par;
i_par == q_par'Dot;
end architecture level0;
The gyroscope (Fig. 7) based on MEMs works with a similar principle: a seismic mass
and interdigitated structures. These are modeled through capacitors with value changing
on plates movement. Clearly in this case a disc inertial moment must be described, so the
equation is no more the 1DOF one, but describes the equilibrium of a rotation for a
system subjected to an angular acceleration. In the case of a MEMs based gyroscope the
moment of a disc is to be described, as the seismic mass is a disc:
1
I= mr
2
Fig.7. Block diagram of the gyroscope.

Other equations are similar to the accelerometer ones, even though translations are
substituted by rotations. In the following the related VHDL-AMS code is shown.
entity gyroscope is
generic (
constant m : MASS := 0.0; constant r : REAL := 0.0;
constant beta : DAMPING := 0.0; constant d0 : ANGLE := 0.0;
constant a : REAL := 0.0
);
port (
terminal rot_mass_gyro : ROTATIONAL;
terminal e_g_up, e_g_mid, e_g_down : ELECTRICAL
);
end gyroscope;
architecture level0 of gyroscope is
quantity theta across C through rotation_mass_gyro to rotational_ref;

quantity I : MMOMENT_I;
quantity v_up across i_up through e_g_up to e_g_mid;
quantity v_down across i_down through e_g_down to e_g_mid;
quantity q_up, q_down : CHARGE;
quantity d_up, d_down : ANGLE;
quantity cap1, cap2 : CAPACITANCE;
begin
-- Disc inertial moment
I == 0.5*m*r**2;
-- Differential equation
C == -I*(theta'Dot'Dot) - beta*(theta'Dot);
-- Relative displacement
d_up == d0 + theta;
d_down == d0 - theta;
-- Capacitance variation
cap1 == a*EPS0/(d0+theta);
cap2 == a*EPS0/(d0-theta);
-- Generated current
q_up == cap1 * v_up;
q_down == cap2 * v_down;
i_up == q_up'Dot;
i_down == q_down'Dot;
end architecture level0;
It is interesting to observe internal waveform behavior. First, in Fig.8 an example of the

voltage used to power the interdigitated fingers of both the accelerometer and the
gyroscope is shown. Typically this is a sine wave with a 300mV peak voltage and with
1MHz frequency.

Fig.8 Voltage to the MEMs interdigitated plates.
Now, as an example, we can apply a linear or rotational acceleration to the system, with a
low frequency sinusoidal wave. As a result, the two capacitance behaves like in Fig. 9:
initially they start from the same value, then, due to the movement of one of the
interdigitated set of fingers capacitances show a specular, sinusoidal behaviour.
Fig.9 Capacitance variation in the gyroscope if the input force is a sine wave.

The gyroscope output is in Fig. 10. It is modulated and the "internal" variation is not
visible due to the high frequency of the reference sine wave (1MHz).
Fig. 10. Gyroscope output if a sine wave acceleration is at the input.
This output is then digitized by ADC's and processed by the output block, which allows
to transform the sensor information into information on the whole vehicle behavior.
RESULTS
Simulations in normal conditions. The first test reported here is the analysis of the
effect of a step as a input displacement. The sensor receives in this case at the input a
pulse with the behavior in Fig. 11. In the figure, accelerometer output is shown, and,
finally, the correspondent vertical displacement behaves as expected.

Fig. 11. Simulation of AHRS when the vehicle is subjected to a step (vertical).
The second analysis is related to a vehicle which is in motion and crosses a bump (Fig.
12). The signal at input is a sine wave which emulates the movement to which the vehicle
is subjected: at the beginning a compression on suspensions is present when the car
mounts on the bump, and later, when the vehicle goes down, suspensions are extended.
The modulated output Z of the accelerometer is a digital value (not reported in figure).
This is processed by the data processing block and shows the vertical movement of the
vehicle, which clearly resembles a bump.
The third analysis shows the accelerometer longitudinal output when the vehicle moves
with a constant speed. This simulation (Fig. 13) differs from a case in which the vehicle
is in static conditions, as the accelerometer mass is in a constant condition, but is not in
its relaxed position. Input is in this case a constant acceleration on the accelerometer (x
axis). The output is constantly modulated and the processed information gives a correct
vehicle behavior, as it moves with a uniform speed along the x axis.
Another interesting analysis is related to the case of the vehicle which moves along a
bend with a constant speed. In this case several are the sensors interested. One is exactly
identical to the one obtained in previous case (Fig. 13), as the vehicle has a uniform
speed, so not reported for sake of brevity.

Fig. 12. Simulation of AHRS when the vehicle is subjected to a bump (vertical).
Fig13. Simulation of AHRS when the vehicle moves with constant speed (horizontal).

Another output is the lateral accelerometer (Fig. 14). In this example the input is a sine
wave, but in reality it will be less regular. However, the output correctly shows an initial
increasing value of the lateral movement, and, after, a decreasing movement when the
vehicle is exiting from the curve. The gyroscope (Fig. 15) detects a yawing due to the
curve. The output after processing is the rotation angle around Z. If the speed were lower,
this angle would have been approximately zero. In case of a higher speed the vehicle
could have problems and a spin could be sensed, as shown in the following simulation
run. Another output (Fig. 16) is still to be analyzed for this simulation: the angle around x
axis. A roll is in fact detected, and gives information to the driver on the vehicle stability.
The angle is zero at the beginning, then increases reaching an angle of 1.5 degree (speed
is 50Km/h), so the curve is a big one.
Fig. 14. AHRS simulation when the vehicle moves with constant speed on a curve.
Lateral accelerometer output.

Gyroscope vertical output.
Gyroscope longitudinal output.

The simulations reported up to this point were related to standard situations. In the
following we are showing what happens when the vehicle is in a critical situation and
how parasitic components included in the models may interfere with data available on
vehicle behavior.
Vehicle loosing stability. We want to analyze here how a spin can be detected. We
show here only the starting part of the vehicle spin (the whole simulation requires very
long computation time). This situation is complex to be managed, as sensors axes are in
motion and might give instantaneous values outside their dynamic range. Moreover a
natural overflow exists, i.e. when the two masses of the interdigitated structures become
in contact.
The vehicle behaviour is simulated with a speed of 60Km/h. The longitudinal output of
the accelerometer behaves exactly as in Fig. 13, except for the numerical values related to
a higher speed (the graph is not reported for sake of brevity).
Fig. 17. AHRS simulation when the vehicle moves with high speed on a curve and is
loosing stability. Accelerometer lateral output.

The lateral output (Fig. 17), which gives information on the vehicle displacement shows a
parabolic behavior due to the input acceleration. This lateral acceleration will force the
vehicle to a spin. The gyroscope mounted on Z axis (Fig. 18) senses a constant torque
(simplification of a real condition in which this torque will be moving around this value).
The consequence is an increasing value of the yawing angle, that will cause a spin on the
vehicle. In this simulation the yaw angle increases of 17 degrees at each second. Clearly,
in case a ESP system is mounted on board this situation will be corrected.
Fig. 18. AHRS simulation when the vehicle moves with high speed on a curve and is
loosing stability. Gyroscope vertical output.
Simulation in presence of errors. Simulation conditions were considered ideal, up

to this point. Anyway, several causes of errors could be present in the real world, e.g.
displacements when mounting sensors on the board and the board on the vehicle, non-
idealities of sensors and other components, etc.
In order to demonstrate that using our methodology it is possible to take into account
these types of problems we inserted a parasitic capacitance to the outputs of the two
sensors. As it is currently difficult to obtain real values from the constructors, and also
because in this case the aim is to show the methodology potentiality and not to find
specific values that cause errors, we have chosen two capacitance values: a small one and
a big one. We show here the case of a vehicle over a bump, as in the previous (Fig. 12)
ideal case. Fig. 19 shows the behavior of the processed output (a vertical displacement)
with and without parasitic capacitance added to the model. In this case the capacitance is
small and corresponds to 20fF. The behavior is similar, but the values are different. In the
error case the displacement is approximately 7mm smaller. It means that in this case, the
values read at the output correspond to a bump smaller than the real one.

Fig. 20 refers to an identical simulation, but the capacitance is now 200pF. The processed
output is totally wrong. In this case a calibration would be needed and an accurate data
processing is mandatory to reduce the error.
Fig. 19. AHRS simulation when the vehicle crosses a bump and a non-ideality (small) is
added to the sensor model. Accelerometer vertical output.
Fig. 20. AHRS simulation when the vehicle crosses a bump and a non-ideality (big) is
added to the sensor model. Accelerometer vertical output.

CONCLUSIONS
This chapter presented a model of an AHRS platform described using VHDL-AMS. The
system is simulated considering real external conditions and cases of normal and critical
behavior were analyzed. The description also allows to insert errors in the sensors
description, and thus to analyze the impact on the platform performance. Besides the
novelty of the inertial platform description itself, we demonstrated how this methodology,
which exploits hierarchy, multidisciplinary and multiresolution capabilities, can be
effectively adopted during the design phase in order to asses system limits and to find the
best corrections.
ACKNOWLEDGEMENTS
We would like to aknowledge Paolo Barbero for his contribution in applying this
methodology.
REFERENCES
1076.1, I. S. (1999). VHDL Analog and Mixed-Signal Extensions, IEEE Std. 1076.1.
Adiprawita1 W, Suwandi Ahmad1 A., Sembiring J., "Development of AHRS (Attitude and
Heading Reference System) for Autonomous UAV (Unmanned Aerial Vehicle)
", Proceedings of Electrical Engineering and Informatics, Indonesia 2007
Ashenden P., Peterson G. and Teegarden D., DE (2002). "The System Designer's Guide to
VHDL-AMS, Elsevier.
Barbero P., "Simulation and modeling of an automotive positioning system", Politecnico di

Torino, Thesis, 2006.
Boussetta H., Marzencki M, S. B. and Soudani, A. (2010). "Efficient physical modeling of

mems energy harvesting devices with vhdl-ams", IEEE SENSORS JOURNAL Vol. 10(No.
9): 1427 1437.
Casier H, P. M. and Appeltans, K. (1998). "Technology considerations for automotive",

Proceedings of IEEE International Design and Test Conference (DATE98, Paris, France.
Casu M.R., Graziano M., Crepaldi M. and Zamboni, M. (2007). "An effective ams top-
down methodology applied to the design of a mixed-signal uwb system-on-chip",
Proceedings of IEEE Design Automation and Test Conference, Nice, France.
Casu M.R, Crepaldi M.. and Graziano, M. (2008). "A vhdl-ams simulation environment for
an uwb impulse radio transceiver", IEEE TRANSACTIONS ON CIRCUITS AND
SYSTEMSI: REGULAR PAPERS, Vol.55(No. 5): 1368 1381.

Christen, E. & Bakalar, K. (1999). "Vhdl-amsa hardware description language for analog
and mixed-signal application"s, IEEE Transactions on Circuits and SystemsPart II:
Analog and Digital Signal Processing Vol. 46(No. 10): 12631272.
Cordoba M., Attitude and heading refernce system I-AHRS for the EFIGENIA autonomous
unmanned aerial vehicles UAV based on MEMS sensor and a neural network strategy for
attitude estimation, Proceedins of IEEE Mediterranean Conference on Control and
Automation, 2007.
Gao F, Blunier B., A. M. & Moudni, A. E. (2010). "A multiphysic dynamic 1-d model of a
protonexchange-membrane fuel-cell stack for real-time simulation", IEEE TRANSACTIONS
ON INDUSTRIAL ELECTRONICS Vol.57(No. 6): 18531864.
Friedman, J. (2005). "Matlab/simulink for automotive system design", Proceedings of

Design and Test Conference (DATE05), Munich, Germany.
Gebre-Egziabher D., R.C. Hayward and J. D. Powel, USA. A Low-Cost GPS/Inertial Attitude
Heading Reference System (AHRS) for General Aviation Applications, Proceedins of
IEEE Position, Location and navigation Symposium, Palm Springs, CA, April 1998.
Graziano M, Ruo Roch M. "An Automotive CD-Player Electro-Mechanics FauSimulation

Using VHDL-AMS". J. Electronic Testing 24(6): 539-553 (2008)
Gursoy M., Jahn, S, B. D. and Pelz, G. (2008). "Methodology to predict eme effects in can
bus systems using vhdl-ams", IEEE TRANSACTIONS ON ELECTROMAGNETIC COM-
PATIBILITY Vol. 50(No. 4): 9931002.
Li Y., Dempster A., Li B., Wang J., Rizos C. "A low-cost attitude heading reference system
by
combination of GPS and magnetometers and
MEMS inertial sensors for mobile applications
", Journal of Global Positioning Systems 5(1-2): 88-95, 2006.
Li Y., Efatmaneshnik M., Cole A., Dempster A.G, "Performance Evaluation of AHRS
Kalman Filter for
MojoRTK System
" , International Global Navigation Satellite Systems, 2009.
MentorGraphics (2004). ADVance MS (ADMS) Reference Manual, Mentor Graphics.
Moser, E. & Mittwollen, N. (1998). "Vhdl-ams: The missing link in system design -
experiments with unified modelling in automotive engineering", Proceedings of IEEE
International Design and Test Conference (DATE98), Paris, France.

Nguyen N., P. L. and Naviner, J.-F. (2005). "Vhdl-ams behavioral modelling and simulation
of high-pass delta-sigma modulato"r, Proceedings of IEEE International Behavioral
Modeling and Simulation Conference (BMAS05), San Jos` , California.
Oudinot J., Scotti S, J. R. and Le-clerc, A. (2003). "Full transceiver circuit simulation using
vhdl-ams", Journal of Microwave Engineering pp. 2933.
Pecheux F, C. L. and A.Vachoux (2005). "Vhdl-ams and verilog-ams as alternative

hardware description languages for efficient modeling of multidiscipline systems", IEEE
Transactions on Computer-Aided Design of Integrated Circuits and Systems Vol. 24(No.2):
204225.
Radezki M., "Languages for Embedded Systems and Their Applications" (2009), Springer,
Lecture Notes, in Electrical Engineering.
Ripka P. "Magnetic Sensors and Magnetometers", Artech House Publishers (2001).
Santarini, M. (2006). "Design challenges steer automotive electronics", EDN Magazine.
Wang, L. & Kazmierski, T. (2005a)." Vhdl-ams based genetic optimization of a fuzzy logic
controller for automotive suspension system", Proceedings of IEEE International
Behavioral Modeling and Simulation Conference (BMAS05), San Jos` , California.
Wang, L. & Kazmierski, T. J. (2005b). "Vhdl-ams modeling of an automotive vibration

isolation seating system", Proceedings of IASTED international conference on Signals and
Systems (CSS 2005), Marina Del Rey, USA.
Wen-shu N., Liao-ni W. and Qi L., AHRS base on MEMS-IMU for aircraft model in wire-
driven parallel suspension system, Proceedings of the International Conference on Mechanic
Automation and Control Engineering, August 2010.
Zhang K., Fu L., Fan Y., Design of Strap-Down Attitude and Heading Reference System
Apparatus Based on DSP and FPGA, Journal of Measurement and Control Technology,
2007.

Section 3
Electrical Engines and Environment Protection
Biolubricants and Triboreactive Materials for Automotive Applications

Amaia Igartua, Xana Fdez-Prez, Olatz Areitioaurtena, Rolf Luther, Christian Seyfert,
Jrgen Rausch, Iaki Illarramendi, Michael Berg, Hubert Schultheib, Benjamin Duffau,
Sylvain Plouseau, Mathias Woydt
Design Methodology for Small Brush and Brushless DC Motors

Jrme Cros, Mehdi Taghizadeh Kakhki, Geraldo C.R. Sincero, Carlos A. Martins,
Philippe Viarouge
Development of a Co Sensor for Hydrogen Fuel Cell Powered Vehicles

Prof. Giovanni Neri, G. Leonardi, M. Parthibavarman, V. Hariharan, C. Sekar
Biolubricants and Triboreactive Materials for Automotive
Applications
Amaya Igartua and Xana Fdez-Prez from Tekniker, Rolf Luther, Christian
Seyfert and Jrgen Rausch from Fuchs , Iaki Illarramendi from TARABUSI,
Michael Berg, Hubert Schulthei from IAV, Benjamin Duffau, Sylvain Plouseau
from Renault and Mathias Woydt from Bam
ABSTRACT
The research institution TEKNIKER has coordinated the EUROPEAN Project EREBIO,
were different biodegradable lubricants have been formulated by FUCHS and BAM for
heavy duty engines (GUASCOR), and passenger cars (RENAULT). In the frame of this
article, it has been summarized the results obtained when developing biodegradable
passenger car lubricants in combination with triboreactive materials.
Replacing hydrocarbon-based oils with biodegradable products is one of the ways to
reduce adverse effects on the ecosystem caused by the use of lubricants. The application
of low sulphur, ash and phosphorous (lowSAP) ester- or polyglycol based oils, intended
for passenger car engine lubricants as substitutes for hydrocarbon-based oils, required the
preparation of a composition of lubricants with comparable tribological and functional
properties.
The study is focused on passenger car motor oils (PCMO) with reduced metal-organic
additives. This is necessary in order to reduce the ash build-up, in the after treatment
system and therefore improve its efficiency and lifetime. High fuel efficiency and long
drain intervals are requested, as well. To follow a line in a consequent way, these oils
have to be biodegradable and non-toxic to the aqueous environment according to the
directive EC/1999/45, coherent with other international standard. In a modern diesel or
Gasoline engine, the engine oils has to fulfil quite a number of different functions, such
as lubricating and cooling the system, wear protection, soot and particle handling with
less deposit tendency and so on.
In the paper a study of the biodegradability, toxicity and the tribological properties has
been carried out for new developed prototype engine bio-oils. Also, some different
plasma sprayed triboreactive coatings have been deposited on cast iron piston rings,
being studied also their tribological properties.
Finally, the behaviour of the new bio-oils selected and plasma sprayed triboreactive
coatings on piston rings have been screened in a real engine.

INTRODUCTION
Recently the book Biodegradable Lubricants has been edited by TEKNIKER. It

summarizes the different research carried by partners from the Virtual Tribological
Institute in relation to biodegradable lubricants (Igartua A, Barriga J, Aranzabe A, 2005).
For this study, the recent literature published by BAM (Schmidt R, Klingenberg G, Woydt
M. 2006 and Desplanches G, Criqui B, Linneman T, Woydt M. , 2006) and the international
literature, has been analyzed (Ramadhas AS, Muraleedharan C, Jayaraj S, 2003 and Gmble
RJ, Priest M, Taylor CM, 2003).
EXPERIMENTAL PART
The candidate oils based on a blend of hydrocarbons with esters (HCE) named EPC-48
and EPC-49, are proprietary by Fuchs Petrolub. A comparison of the project
specifications and the candidate oils data is given in Table 1.
Table 1.: Candidate engine oils for the PCMO specification
PCMO-Specifications epc-48 epc-49

SAE grade 0W-30 or 5W-30 0W-20
Viscosity kin.V. 40C < 55mm/s 44mm/s
kin.V. 100C > 9.3mm/s 8.8mm/s
CCS -25C < 2900mPas 1660mPas
HTHS 150 C 2,9 - 3,1mPas 2.9mPas
Shear stability (Bosch) stay in grade Excellent
Pour point < -35C -45C
P < 800ppm 560ppm 560ppm
Elements S < 2000ppm 1800ppm 1800ppm
Zn < 500ppm free Free
sulphated ash < 0,5% 0,5% 0,5%
Noack evaporation loss < 13% 5,2 %
ICOT oxidation stability 140 h pass O.K.
Both new oils are HCE based on the same base fluid composition but differ in
additivation. Due to their high viscosity index, the desired high temperature, high shear
viscosity (HTHS) at 150C of 2.9 mPas is reached at significantly lower viscosities at
100, 40 and at -25C. This offers the potential for improved fuel efficiency in city
driving. As explained earlier, the major challenge in the formulation of the new oils is the
reduction in crucial additive elements. Sulphur and phosphorous are the elements found
in most anti-wear agents, while the metal-organic compounds constitute important anti-

oxidants and engine cleanliness additives. All three additive groups can have negative
influence on the after-treatment system; therefore, the so called lowSAP (reduced
sulphur, ash and phosphorous) oils are required.
Fig. 1. shows a comparison between the project reference lubricant (5W-30) and the new
bio-oils. The ash content is reduced to one third, phosphorous and sulphur to less than
50% of todays product. Despite this strong decrease in metal-organic compounds, the
deposit formation tendency as screened with the panel coker test according to Wolf
remains as low as the reference oil (Fig. 2). This could be reached by using a new ashless
EP/AW additive technology. However, the influence of the concentration of metal-
organic compounds remains visible even with newest technologies. This is illustrated in
Fig. 3.
Fig. 1. Comparison of additive elements of reference oil and EPC-48 and EPC-49
Fig. 2. Wolf comparison between reference oil and EPC-48.

Fig. 3. Results of Wolf test showing the dependency of deposit formation from the
sulphated ash content of the engine oil, even with most modern D&D (detergent &
dispersant) additives. The oils are identical except for their D&D additive content.
The second concern while using lowSAP oils is wear protection. To check this
performance, screening tests have been performed by the project. Fig. 4 shows results
obtained with the SRVs test rig. On the left hand side, wear scars according to DIN
51834-2/ASTM D6425-05 are displayed. This is a 2 h test at constant load. Here, both
new engine oils show slightly smaller wear scars. On the right hand side, results
according to ASTM 5706-05 are shown. In this SRVs test, the load is increased in fixed
intervals until seizure occurs. Higher load corresponds to a better protection at highest
load spikes. The reference oil and both new candidate oils easily supersede the values
typically found in engines.

Fig. 4. SRV test results obtained for the reference oil and the new EPC-48 and EPC-49
One of the activities from BAM was related to the blending and characterization of the
polyglycol lubricants, which respect an ash-, metal- and polymer-free philosophy.
Table 2.: Properties of polyglycol-based lubricant tested.
ICOT Life Aquatic toxicity

HTHS OECD
TAN or NOAC 40 algae daphnia fish
Lubricant 150C 301F
40C+ K (%) (mm2/s) OECD (mg/l)
(mPas) (%)
100% 201 202
PPG32-2+2.6%
>96h 5.3 34.3 2.87 70.5 >100 600
Phopani
PAG D46-4
>96h 6.2 49.6 3.65 85 >100 >100
+2,6% Phopani
Water soluble PAG (polyalkylene glycols) and oil soluble PPG (polypropylene glycols)
have been considered. Oxidative evaporation losses and the resistance against oxidation
of PPG have been evaluated using the French ICOT (iron catalysed oxidation test) test,
which is standardised in GFC Lu 36T 03. For example, 96 h in the ICOT test corresponds
to a potential of drain of up to 430 km. Some polyglycol formulations exceeded a life
time of 130 h underlining a potential for exceeding ACEA A5/B5 or a drain interval of 40
km. The data in Table 2 underline, that proprietary made base oils and additive packages
can meet the bionotox-criteria.

BIODEGRADABILITY AND TOXICITY TESTS
Biodegradability and toxicity tests were performed by TEKNIKER in order to assess the
biodegradability and toxicity of the new lubricants.
Biodegradability
Table 3 and Fig. 5 show results obtained with new biolubricants. The reference substance
is an internal reference for OECD 301F biodegradation test. The passenger car motor oil
(PCMO) reference was not readily biodegradable. The new PCMO developed pass the
limit of biodegradation (60% in 28 days).
Table 3.: Result of EPC-48, EPC-49, PPG-32 and PAG-46 samples.
Time % Degradation
Ref. EPC48 EPC48 EPC 49 EPC 49

Day PPG-32 PAG-46
Sust. (a) (b) (a) (b)
0 0 0 0 0 0 0 0
7 90.96 45.11 37.46 40.76 41.72 12.3 21.51
14 90.38 70.34 55.05 57.65 61.63 44.4 41.77
21 89.50 81.42 66.13 66.10 74.06 75.4 62.6
28 89.50 87.54 78.36 76.04 84.00 79.3 80.29
Fully formulated motor oils
Fig. 5. Result of EPC-48, EPC-49, PPG-32 and PAG-46 samples. Fully formulated motor
oils.

Aquatic Toxicity
The following basic level test are proposed and included in the EU hazard assessment of
substances: Daphnia acute immobilization test (OECD 202; Table 4).
Table 4. OECD 202, Daphnia Magna Toxicity Test Result
Lubricant *EC50 (mg/l) Classification
Reference PCMO Not tested -
EPC-48 EL50>100 Not harmful to aquatic organisms
EPC-49 EL50>100 Not harmful to aquatic organisms
PPG32-2 EL50>100 Not harmful to aquatic organisms
PAG-46 EL50>100 Not harmful to aquatic organisms
*EC50: Effective Concentration. The EC50 value means the mean effective
concentration of a substance that produces a particular, previously defined behaviour in
50% of the organisms of a test series.
TRIBOLOGICAL TESTS
Tribological tests have been performed by TEKNIKER to simulate different engine

critical components like piston rings and journal bearing simulations.
Piston Ring Cylinder Liner Simulation
To study these new oils have been tested different piston rings coated with triboreactive
powders deposited by Plasma spray by CIE TARABUSI against cast iron cylinder liner,
using the PCMO oil as reference. In the next figure, the configuration to carry out the
simulation tests developed by TEKNIKER is shown (Fig. 6).

Fig.6. Cylinder linerPiston ring configuration test.
The test conditions were: time: 41 min, load: 50/2000N, frequency: 50 Hz, stroke: 3mm,
temperature: 200C. The load is increased in fixed intervals until seizure occurs. Higher
load corresponds to a better protection at highest load spikes. Table 5 presents the results
obtained. These two developed coatings have better extreme pressure behaviour than the
reference Chrome plated piston ring.
Also, the new oils developed (EPC-48, EPC-49, PPG-32 and PAG-46) have similar or
better behaviour than the PCMO reference oils. It seems that the combination
triboreactive coating/polar lubricant increase the load carrying capacity and the extreme
pressure properties.

Table 5. Cylinder liner-Piston ring Test Results
Extreme Pressure Conditions
The highest
Piston rings Oil Time at which
friction load at
failure occurs Type of failure*
which no failure
(minutes)
occurs (N)
Chrome plated 28 1300 Stroke<0,3 mm
TinO2n-1 Ref. 36 1800 Stroke<0,3 mm
Tin-2Cr2O2n-1 29 1400 Stroke<0,3 mm

EPC-
TinO2n-1 32 1600 Stroke<0,3 mm
48
Tin-2Cr2O2n-1 41 2000 No failure
32 1500 Stroke<0,3 mm
Chrome plated
31 1500 Stroke<0,3 mm
EPC-
36 1800 Stroke<0,3 mm
49
TinO2n-1
38 1900 Stroke<0,3 mm

PPG
TinO2n-1 32-2 35 1700 Stroke<0,3 mm
3.2%
Tin-2Cr2O2n-1 34 1600 Stroke<0,3 mm

PAG
TinO2n-1 25 1200 Stroke<0,3 mm
46
*The failure is considered when the first micro-welding appears.

Journal Bearing Simulation
It has been selected the EPC-49 new bio-oil and we have chosen two coated discs
(Typical sheet steel of bearing (50CrV4) coated by TEKNIKER with TiCN+DLC
comparing with a reference material AlSn20) which tested against the steel
countermaterial (50CrV4).
The test configuration selected by TEKNIKER was a standard thrust washer test
configuration from FALEX. The tests conditions were: pressure: 0.5MPa (324 N), initial
temperature: ambient (1525C), speed: 1750 rpm (4.25 m/s) (Fig. 7).
Fig.7. Thrust washer test for journal bearing simulation
Figs.8. and 9. show the results obtained. The new bio lubricant has better friction
behaviour than the reference motor oil 5W-30 with both coated and uncoated material.
Tribological tests show a good extreme pressure and friction behavior in two different
simulations tests of the critical automotive engine components tested (piston ring and
journal bearing simulation).

Fig.8. Results obtained: 50CrV4//typical sheet steel of bearing coated with TiCN+DLC.

Fig.9. Results obtained: 50CrV4//typical sheet steel of bearing coated with AlSn20,
reference
ENGINE TESTS
Scuffing Tests
The piston ring component engine tests have been performed by CIE Tarabusi, SL in a
2.7 L turbo-charged 4-stroke diesel engine, using as reference the Mo-based plasma
coating PL72 (reference used by Tarabusi).
The test conditions and parameters were fixed for three different type of tests: (a) Scuff
test to determine the scuffing resistance of the coating at high temperature and pressure
with minimum component clearance, (b) Hot test, long run at high power and high speed
to set the reference values for wear and (c) Cycle test between low and high thermal

conditions to determine the thermal fatigue and adherence resistance of the developed
coatings. The components were measured before and after the test.
In Fig.10. the difference between measurements are presented. These values are a
combination of the wear and deformation that the components have suffered during the
test. The main conclusions of the tests are summarized as follows:
Good wear and anti-scuffing properties of Titanium Oxide followed by the

(Ti,Mo)(C,N)+23NiMo coatings.
Wear resistance of TinO2n_1 coating is similar to standard Mo-based coating
for both standard and triboreactive oil.
The best tribological properties were found for the Titanium Chromium Oxide coating
but it was detached from the iron substrate during the engine tests.
Wear results with triboreactive oil EPC-49 are similar to the standard oil, but oil
consumption results were reduced by 45%.
Fig.10. Testing of piston rings in a turbo diesel engine testing arrangement.
Plain Bearing and Engine Tests
IAV GmbH carried out primary tests on a real size plain bearing model test bench and
engine tests in a special torque loss test bench. The aim of the work was to evaluate new
bearing materials, a new piston ring coating and new EPC49 lubricant under near-
realistic boundary conditions with regard to their functional reliability, and to make
statements on their friction behaviour. As new bearing materials, coatings, TiOx and
TiCN+DLC were compared with a reference bearing (AlSn20). As it was predicted in the
simulation tests the best results were obtained with the reference bearing. The bearing

showed the running behaviour corresponding to the present application. Unfortunately,
the results obtained with the new coatings applied on the bearing shells are not
satisfactory. Design modifications are indispensable to improve adaptability or to avoid
form deviations. As new piston ring coating, a TiOx-coated first compression ring was
tested by IAV being compared to a conventional production piston ring. No friction
reduction could be verified with the TiOx-coated piston rings.
After approximately 25h running-in time of the piston rings, the friction torque was
similar to the production piston rings. It was possible to reduce the friction of the
examined engine with the developed oil, to a distinct degree especially at the operating
points:
Low temperatures and high engine speed.
High temperatures and low engine speed.
The measured calculated reduction in friction coefficient corresponded to a reduction of

fuel consumption of 2%. During the investigations, the test engine ran for all together
more than 100h. On examining the engine after the tests, no serious signs of wear were
found in the engine components.
Table 6.: The piston ring component engine tests results.
Radial
Radial ring
Time Coat Blow By Oil Consume cylinder Pass/no
Oil Test Type wear Note
(h) composition 55.7l/min (gr/h) wear pass
(/100h)
(/100h)
102.5 Scuff/
Mo+Mo2C+
STD duration 55.7 48.3 25 10.1 Pass Ref.
NiCr
100 Cycles
102.5 Scuff/
STD TinO2n-1 duration 45.6 42.2 0 13 Pass OK
100 Cycles
102.5 Scuff/
EPC- Mo+Mo2C+
duration 49.4 20.9 2.1 14.7 Pass OK
49 NiCr
100 Cycles
102.5 Scuff/
EPC-
TinO2n-1 duration 50.6 20.5 2.2 14.7 Pass OK
49
100 Cycles

Fig.11. Friction moment at assembly level 1 (complete engine 1.9 L TD, series) at 135C
Fig.12. Crank case+piston power group at 135C
CONCLUSIONS
The new passenger car motor oils developed (EPC-48, 49, PPG-32, PAG-46) are
biodegradable according to OECD 301F and non-toxic according to OECD 202 Daphnia
Magna test.
Piston ring/cylinder liner simulation tests developed by TEKNIKER, showed
better extreme pressure properties for Tin 2Cr2O2n_1 and TinO2n_1 (see
Table 5). The results were confirmed by BAM. Both coatings Tin-2Cr2O2n_1
and TinO2n_1were deposited by Plasma by CIE Tarabusi, showing good

tribological properties in engine tests (see Fig. 10), but the coating Tin-
2Cr2O2n_1 was detached in engine test because of their fragility, being
selected the second solution TinO2n_1, potentially more economic and slightly
better in performance than the reference Mo-based coating in combination with
the SAE 5W-30 API CE/SF reference and lowSAP EPC-49 ester-based oil. The
EPC49 presented better extreme pressure properties (see Table 5) at laboratory
simulating tests, reducing also the oil consume until 45% in engine tests (see
Table 6), probably due to the lower NOACK volatility of the oil.
Correlation between journal simulation test (TEKNIKER) and journal bench

tests (IAV) have been found. In this configuration soft/hard alternative
presented better behaviour than hard/ hard alternative. Design modifications are
indispensable to improve adaptability or to improve edge load carrying
capability of the hard/hard-couples. The biolubricant EPC 49 reduced
significantly the torque and temperature (see Figs. 8 and 9) in the journal
simulation tests and also the friction in the journal bearing and engine test from
IAV (see Figs. 11 and 12). In engine test, the alternative engine oils PPG32-2
and hydrocarbon-ester blend EPC-49 displayed a reduction in friction, which
equals at least to a reduction of fuel consumption of 42%.
In conclusion, a new generation of biodegradable and low ash prototype oils

and coatings for Passenger car applications have been developed, being
possible to predict the behaviour of engine tests, simulating their Tribological
behaviour at the laboratory. The combination of triboreactive coatings with
polar lubricants seems to be synergistic improving the extreme pressure
properties and wear resistance. The great advantage of the new developed Low
SAP oils is that they are more compatible with the after treatment system due to
the lower content of phosphorous, sulphur and ash. Additionally, they are
biodegradable and non-toxic.
ACKNOWLEDGEMENTS
The authors express their gratitude to the European Union for the financial support given
to this work through the GROWTH Project no. GRD2-2001-50119, Contract no. G3RD-
CT-2002- 00796-EREBIO and the Spanish Minister of Science and Technology reference
MAT2002-11137-E. The authors also would like to thank all the remaining partners from
EREBIO project FALEX, CRSA, GUASCOR, FERESPE, BRITO, INEGI for giving
their approval for this publication.
REFERENCES
Igartua A, Barriga J, Aranzabe A (2005). Biodegradable Lubricants Virtual Tribology

Institute Edition, ISBN 83-70204-418-X.

Schmidt R, Klingenberg G, Woydt M. (2006) Tribological, thermophysical and
viscosinetric properties of lubricants interacting with triboactive materials. BAM
Research Report 277;
Fitamen E, Tiquet L, Woydt M. (2006) In: ASTM D02 Symposium on Automotive

LubricantsTesting and Additive Development.
Desplanches G, Criqui B, Linneman T, Woydt M (2006). Plenary paper al 15th

International Colloquium Tribology, TAE Esslingen.
Rakopoulos CD, Antonopoulos KA, Rakopoulos DC (2006). Energy conversion and

Management 47(1112):155073.
Ramadhas AS, Muraleedharan C, Jayaraj S. (2003) Renewable Energy 1789800.
Miura M. (2004) Journal of Japanese Society of Tribologists 49(10):7938.
Ulosoy Y, Tekin Y, Centinkaya M. (2004) Energy Sources 26(10):92732.
Megahed OA, Abada RI, Nabil D. (2004) Energy Sources 26(2):11926.
Tanabe H. (2003) Journal of Japanese Society of Tribologists 48(6):4426.
Hashimoto T. (2003) Journal of Japanese Society of Tribologists 48(4):27882.
Chen CI. (2003) Tribology Letters 14(2):8390.
Gmble RJ, Priest M, Taylor CM. (2003) Tribology letters 14(2):14756.

Design Methodology for Small Brush and Brushless DC
Motors
Jrme Cros, Mehdi Taghizadeh Kakhki, Geraldo C.R. Sincero,
Carlos A. Martins, Philippe Viarouge
INTRODUCTION
Manufacturers of components and sub-systems for large scale applications are constantly
improving the quality of their products in order to ensure customer satisfaction. They are
facing severe competition constraints and this is particularly true for automotive industry.
The offer of new services and functionalities goes well together with the increasing use of
electronics and electrical actuators which provide auxiliary functions in a vehicle,
improving the functional performances.
Given this context, and the constant need for improving the electrical machines, the
present chapter focuses on the design methodology for brushed or brushless DC machines
as the two mostly used motor types in small accessories for vehicles. This methodology is
also applicable to many other machine types. We discuss various winding configurations
for the machine armature and the advantages of new motor designs using a concentrated
winding. We present a method for the selection of an efficient motor and an analytical
dimensioning process using a non-linear constrained optimization approach. Finally, we
discuss the application of finite element and time-simulation methods for the validation
of the optimal solution.
SMALL ELECTRIC MOTORS IN MODERN VEHICLES
An electrical actuator is a device designed for moving or controlling another mechanism

or system. Table 1 shows a list of automotive accessories using small power actuators,
(Cho and Johnston, 1999). Small rotating electric motors are an important part of the
electrical system and are used in several feature accessories in a modern car. Over 100
motors are used in a well equipped luxury vehicle (Thiemer, 2001). These machines have
very different requirements in terms of power, speed, torque, volume, form and size.
The energy conversion from electrical to mechanical in a rotating electrical motor is the
result of an interaction between two magnetic fields, one created by permanent magnets
(PM) or electromagnets and the other generated by current-carrying conductors (armature

winding). The number of magnetic poles in the magnetic field is referred to as machine
poles. The current-carrying conductors are mounted in the armature core openings which
are referred to as armature slots.
Table 1. List of automotive accessories using small power actuators (Cho and Johnston,
1999).
starter motor, alternator, radiator motor-fan, air conditioning
Power train compressor drive, idle speed control, engine throttle control,
transmission shifter, electrically variable transmission, engine coolant
pump motor, electrical valve, and ECR actuators.
electric power steering system, electro-hydraulic power steering, ABS
Chassis systems, brake-by-wire actuators, active suspension actuator, and 2-4
wheels drive actuator.
windshield wipers, window lifts, seat adjuster, seat vibrators, sunroof
Body actuators, sliding door closers, door lock mechanisms, headlamp
adjuster, mirror adjusters, steering column adjuster, HVAC blower,
cruise control, headlight wiper motors, power antenna, headlight doors,
trunk closer, and auto-leveling system.
The two mostly used motor types in small accessories are PM brushed dc motor and PM
brushless dc motor. In a PM brushed dc machine, the permanent magnets are mounted on
the stationary part of the machine (stator) and the armature is the rotating part of the
machine (rotor). DC voltage is applied to the commutator copper segments through
carbon brushes pressing against these segments (Hamdi, 1994). The commutator
segments are mounted on a cylinder on the rotor shaft and are insulated from each other
and also from the rotor shaft. The commutator segments are wired to the ends of
the armature coils. In fact, the set formed by the commutator and the brush assembly acts
as a mechanical rotary switch. This allows reversing the coil current in the rotating
armature maintaining the torque production. Fig. 1 shows a brushed dc machine used as a
radiator cooling motor fan.
In a PM dc brushless machine, the permanent magnets are mounted on the rotor and the
armature is stationary. The armature coils are fed with a power electronic converter and
the coil current waveform must be synchronized with the rotor position. These motors are
often fed with square-wave currents or voltages when a simplified control system is
required. However, similar motors fed with sinusoidal currents and voltages are also
becoming increasingly popular since the cost of electronics (including microcontrollers)
is rapidly decreasing. Employing a sinusoidal current waveform is more advantageous
since it generates less noise and vibration. However, it requires several current sensors
and increases the complexity of the control system.
The power converter acts, therefore, as an electronic commutator and is generally

integrated inside the motor. It avoids the problems associated with a mechanical brush-
commutator assembly including the mechanical wear and EMI compatibility (Torrey and
Kokernak, 2002) and results in a more efficient and reliable machine. The cost of

production for PM brushless machine is higher compared to the brushed dc machine but
it allows for variable speed operation without an additional electronic converter. It is
usually used in more expensive luxury cars. Fig. 2 shows a 5-phase brushless dc machine
with an external rotor for a radiator cooling motor fan.
Fig.1. Brushed DC motor for a radiator cooling motor fan
Brush assembly
Armature
(rotor)
Permanent magnet poles

(stator)
Commutator
Fig.2. Brushless DC motor for a radiator cooling motor fan
Armature Permanent magnet poles

(stator) (rotor)
Power electronic
converter

Table 2 and 3 show examples of PM brushed dc motors and PM brushless dc motors for
different applications in vehicles.
Table 2. Structure of PM dc motors applied on some of vehicle accessories

Electro- Electro- Heater
Radiator Wiper Starter
Application hydraulic hydraulic blower
Fan Motor
ABS Power Fan
Poles 4 2 4 2 2 6
Slots 20 12 21 12 10 29
Brushes 4 2 4 2 2 2
2565 2000 1820 1800
Speed 1000 rpm 2000 rpm
rpm rpm rpm rpm
500 W
Power 150 W 500 W 120 W 120 W 1.2 kW
1.5 kW
Input
12 V 12 V 12 V 12 V 12 V 12 V
Voltage
Table 3. Structure of PM brushless dc motors applied on some vehicle accessories

Electro-
Radiator Radiator Heater
hydraulic Water
Application Fan Fan blower
Power Pump
Fan
Steering
Poles 4 6 4 6 4
Slots 20 9 8 9 12
Number of
5 3 2 3 3
Phases
1800 3500 5000
Speed 2500 rpm 2000 rpm
rpm rpm rpm
Power 250 W 350 W 120 W 1500 W 500 W
Dc Bus
12 V 12 V 12 V 42 V 42 V
Voltage
Trends in Small Motor Design for Automotive Applications
In the current economic context and due to severe competition, all kinds of electrical
machines should be constantly improved to reduce their costs and improve their
performance. This evolution has become possible thanks to more efficient and precise
calculation methods which allow the designer to further elaborate the optimization of the
structures. This is also due to technological advancements in materials, electronics,
manufacturing techniques and new technical specifications that lead to new developments
and new designs.
For instance, the 42V Powernet voltage modification has important consequences on
existing electrical equipment, mainly for brushed dc motors present in many systems and

accessories. The higher input voltage may lead to current commutation problems,
electromagnetic interferences and, even, ring fire on brushed dc motors, among other
problems (Thiemer, 2001; Torrey and Kokernak, 2002; Hamdi, 1994). In this context,
many existing motors need to be improved. It is also interesting to evaluate new design
solutions that might be more appropriate for the new electrical system specifications.
Integration of the equipment in a new car design adds new constraints and should meet
new requirements. Designers will have to pay attention to a number of aspects such as
frame size, cost, noise, lifetime, ambient conditions, and quantities which vary from one
vehicle model to another.
New designs of electrical motors can benefit from new manufacturing techniques, for
example, new winding methods which reduce the copper volume and copper losses in
significant proportions.
The use of new materials may have important consequences on the machine structure.
For instance, the design of electromagnetic devices with soft magnet composite (SMC)
materials can offer several advantages over conventional laminated materials. It can
facilitate the implementation of new structures with fewer parts, reduced size and weight
(Hultman and Jack, 2003).
DESIGN METHODOLOGY
The design methodology allows the designer to find a compromise between material
properties, device dimensions, the cost of production and assembling processes, and
application constraints. It begins with the analysis of the application specifications and a
topological research to find a machine structure well-adapted for the given application.
The topological research is, in fact, the process of the pre-selection of machine structures
which have the best potential to meet the requirements of a given application. It allows
the designer to limit the number of solutions before proceeding to a dimensional and
geometrical optimization. Fig. 3 shows a general flowchart for this type of methodology.
Topological Research
This approach relies on the designers knowledge and experience (expert rules) and does
not need any particular Computer Aided Design (CAD) tools other than analytical
models. This approach enables the designer to find geometries of electrical machines
which take advantage of particular properties and performance of magnetic materials or
windings. This helps also in reducing the size of the optimization problem. For example,
Soft Magnetic Composite (SMC) materials may be an interesting choice for the type of
magnetic material, however, they may impose some specific constraints for the motor
geometry in regard to the pressing process. Nevertheless, compared to laminated
material, SMC materials have isotropic properties which allows the design of original
structures of magnetic circuits with different mechanical assemblies. This may improve

the production process. This type of decision can not be easily formulated in an
optimization problem and all possible options should be studied individually. The
selection of a winding configuration, the choice of the number of poles and slots in a
machine, or brush and commutator segment numbers in the case of a brushed dc machine,
are also other examples of parameters to determine in a topological research before
proceeding to a dimensional and geometrical optimization.
Fig.3. Design optimization methodology
Topological Research
Analysis of Selection of the structure
application
Axial or radial air-gap, number of poles,
specifications
number of slots, winding configuration,
etc.
Global optimization of a given structure
Dimensioning
Multi-physical modeling Optimization
Prototypes and Tests
Classical Machine Dimensioning Method
In this classical approach, the main dimensions of the machine are derived directly from
the application specifications using simplified analytical formulas and expert factors
determined by experimental data from many years of design experience (Hamdi, 1994).
It may also employ Computer-Aided-Design (CAD) tools like Finite Element (FE)
magnetic field calculations to complete the analysis and to validate the structure. Fig. 4
shows the diagram of a complete CAD environment for electrical machine design. It
should bring together a multitude of coupled physical models to reproduce all the
physical phenomena typically appearing in an electrical machine. It is also possible to
perform several iterations with this CAD procedure. However, this is generally done by
the designer when comparing and analyzing different solutions without using a specific

optimization method. In fact, this classical methodology is often confined to the
comparison of all the solutions resulting from a topological design approach.
CAD Environment
Fig.4. Computer-Aided- with
Design (CAD) coupling
Environment ofcoupling
with physical models
of physical models.
Thermal Modeling Dilation, Mechanical Modeling

Heating transfer dissipation Assembly, parts movement
Iron Losses
Strength,
Vibration
Joule Losses
B(H) curve
Electrical Circuit Modeling Flux Magnetic Modeling

Current & voltage waveforms densities Magnetic flux
Global Optimization
The aim of global optimization is to find the optimal dimensions for a suggested structure
taking into account all kind of application constraints and material properties. The term
Global refers to the fact that this approach takes into account all different physical
phenomena and their interactions. A Global optimization approach needs also physical
models, but unlike the classical approach, it uses the dimensional parameters as design
variables and the application specifications are considered as constraints. The main
challenge is to define an optimization problem with a given objective, variables, and
linear and non-linear constraints (geometrical, electrical, thermal and mechanical). The
choice of an optimization tool and optimization method (such as gradient-based methods,
evolutionary algorithms, etc.) does not depend on the optimization problem. Any method
could be used as far as convergence towards a solution is possible and all constraints are
validated using modeling methods. Both analytical and finite element models could be
used to evaluate the performance of the electrical machine.
Optimization with Analytical Models

Analytical models are widely used in a global optimization process since they allow us to
take into account many physical phenomena and their mutual interactions. They are
simple to implement and fast in terms of simulation time. However, they need
simplifying assumptions which affect their accuracy and limit their utility in applications
with complex geometry and material non-linearities.
Optimization with Analytical Models Corrected with Finite Element Models

Finite Element (FE) models are used to analyze the performance of the electromagnetic
structures before their final realization. They are more precise than analytical models and

are generally used as a reference for the validation of analytical models. Finite element
models are used for field calculation, thermal and mechanical modeling, etc.
Optimization schemes with analytical models corrected by finite element models are
becoming increasingly popular since they benefit from advantages associated with both
analytical models and field calculation tools. In this approach, for each intermediate
optimal design solution, the electrical parameters generated by the analytical model are
compared to those obtained by a 2D or 3D finite element computation (Cros et al., 2008).
If there is a significant difference between these two sets of parameters, correction factors
are applied to the analytical models to improve their accuracy. Figure 5 shows the
flowchart for a design methodology using this approach.
Optimization with Finite Element Models (Direct Method)
Two dimensional (2D) and three dimensional (3D) finite element based methods may
lead to more efficient tools for field calculation, allowing for a more precise evaluation of
the machine characteristics. Currently, however, an iterative optimization process using
finite element models may result in unpractical computation times in the case of electrical
machine design. With the new technological advancements in data processing and
computation speed increase, this approach will be progressively more interesting and will
become possibly predominant in the future.
Fig.5. Flowchart of the design methodology
Load and power specifications

Torque-speed characteristic
and energy requirements
Selection of motor main Slot and pole numbers

parameters Brush and commutator segment numbers
Selection of Position of
a winding phase coils in armature slots
configuration
Iterative geometric 2D FE magnetic computations

design calculation
Analytic electrical Correction factors for electrical

equivalent model model
Performance calculation and Commutator model for time

validation simulation

ARMATURE WINDING CONFIGURATION
In conventional DC motors, three main types of rotor armature windings may be

identified: lap winding, wave winding, and frog-legs winding (Hamdi, 1994). These
windings are made with simple coil elements which are always interleaved. The ratio
between the axial length of the end-windings and the axial length of the armature
magnetic circuit is then relatively high. All these winding types differ primarily on the
method used to connect the terminals of the simple coils to the commutator. For a lap
winding, the number of parallel paths is equal to the number of poles. A wave winding
has only two paths in parallel, regardless of the number of poles. The frog-leg winding
method combines lap winding and a wave winding placed on the same armature, in the
same slots, and connected to the same commutator bars.
Figure 6 presents a conventional lap winding armature of a permanent magnet brush DC

motor used in an automotive application. One can notice that the yoke has a relatively
high number of slots and is made of laminations. The significant copper volume of the
end-windings is typical of a lap winding with a small axial motor length. To avoid the
interleaving of the coils, it is possible to directly wind the armature simple coils around
each tooth of the rotor magnetic circuit (Cros and Viarouge, 2002). This kind of winding
is called a concentrated winding but may be also called a non-superposed winding. As
shown in Fig. 6, this technique considerably reduces the copper volume of the end-
winding, the total copper losses and the total axial length of the motor (Cros and
Viarouge, 2003). In small power applications using permanent magnet motors with
reduced axial length, concentrated windings are progressively replacing other types of
armature windings.
Fig.6. Comparison of armature end-windings
a) lap winding b) concentrated winding

Armatures with concentrated windings have always a small number of slots with wide
openings and large tooth sections. They can be realized with a laminated material but
they are also well adapted to the use of soft magnetic composite (SMC) materials due to
the reduced mechanical constraints during the molding process. With such materials, it is
possible to insert the end-windings in the active part of the rotor magnetic circuit and
expand the tooth tips to perform an axial concentration of the air-gap magnetic flux into

the teeth. Such modification takes advantages of the isotropic properties of magnetic
composite materials and partially compensates its low permeability. This axial insertion
of the end-windings reduces the volume of copper and the total axial length of the motor.
Concentrated Armature Windings of Brushed DC Motors

The concentrated winding technique is too often associated and restricted to windings
with a short pitch, i.e. windings with lower performances than those of classical winding
structures. The concentrated windings with a short pitch are then limited to sub-fractional
power applications with low voltage supply, where the performance in terms of torque to
current ratio is not critical.
In the case of the brushed DC motor, the simplest motor widely used for mirror
adjustment and door latch is using a concentrated winding armature because its
production cost is very low. This motor has 2 poles, 3 armature slots, 2 brushes, 3
commutator bars and a winding with a short pitch of 120 electrical degrees. Fig. 7(left)
shows the diagram of a developed surface of this drum armature. This developed diagram
is made by unrolling the periphery of the armature and commutator into a plane. This
motor, which is well adapted for low voltage and low power applications, presents
several drawbacks:
the current commutation is not efficient, because the number of segments on the
commutator and the number of coils per parallel path are very small (Fig.7, right).
The commutation voltage between two consecutive segments is relatively high r in
important Electro-Magnetic Interferences (EMI).
because the internal voltages in the parallel paths of the winding are unbalanced, the
path currents are different and there is a circulating current in the armature which is
generated by the third harmonic of the no-load emf in each coil. This current
decreases the motor efficiency and increases the noise and the torque ripple.
the winding coefficient associated to a short pitch of 120 electrical degrees is small
and equal to 0.866. The performance in terms of torque to weight ratio and torque
ripple is low.
Fig.7. Diagram of an armature with 3 slots, 2 poles and a single layer concentrated
winding with a short pitch of 120 electrical degrees (3 coils); Diagram of the parallel coil
paths (right)
3.1 2.1 1.1
1.1
V
N S
3.1 2.1 1.1 3.1 2.1 3.1
1 2 3

Fig. 8 (left) shows an evolution of the machine of Fig. 7 by using a multi-layer
concentrated winding. This new winding configuration increases the number of
commutator bars and of armature coils. In Fig. 8 (right), each coil has been identified by
2 numbers. The coils having the same first number (e.g., coils 1.1 and 1.2) are wound
around the same tooth and have the same no-load voltage. These coils are connected to
different bars of the commutator to get coil paths perfectly balanced. The current
commutation in the coils wound on a same tooth is achieved, simultaneously, by different
brushes. Consequently, the multi-layer winding improves the commutation when
compared to a single layer winding because the number of turns per coil and the coil
inductance are reduced for the same operating point (same values of DC supply voltage,
DC current and rated speed). Employing an armature with a multi-layer concentrated
winding is an interesting solution for applications which need more power and higher
voltages.
Fig.8. DC motor with 3 slots, 2 poles and a multi-layer concentrated winding using 6
coils (left); Diagram of the parallel coil paths (right).
3.1, 3.2 2.1, 2.2 1.1, 1.2

1.1
3.2 2.1
3 2 1
V
N
N S
S 3.1
2.1 3.1 1.2 2.2 3.2 1.1 2.1 2.2
1.2
1 2 3 4 5 6
V
Comparison: Lap Winding versus Concentrated Winding
The drive of an automotive electrical motor fan, which is a typical application of the
permanent magnet brushed DC motor, is chosen to compare a conventional lap winding
armature with a multilayer concentrated winding one. The conventional motor has 20
armature slots with 20 simple coils, 4 stator poles and 4 brushes. This type of motor has
been adopted by a lot of manufacturers in the world. The winding of the rotor is
overlapped with a short pitch of 1 to 5. Fig. 9 shows the developed diagram of this
structure with coil connections on the 20 commutator segments. The coils having the
same first number (e.g., coils 1.1 and 1.2) have the same no-load emf and are in phase.
Fig. 10 shows the parallel coil paths for the same machine. Manufacturers sometimes
make simplifications to minimize the cost. For example, the number of brushes can be
reduced while adding equalizer connections on the commutator as shown in Fig. 11.

Fig.9. Diagram of a lap winding machine having 20 rotor slots, 4 stator poles and a lap
winding and a short pitch from 1 to 5
3.1 4.1 5.1 1.2 2.2 3.2 4.2 5.2 1.3 2.3 3.3 4.3 5.3 1.4 2.4 3.4 4.4 5.4 1.1 2.1 3.1
N S N S
5.4 1.1 2.1 3.1 4.1 5.1 1.2 2.2 3.2 4.2 5.2 1.3 2.3 3.3 4.3 5.3 1.4 2.4 3.4 4.4 5.4
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Fig.10. Diagram of parallel coil path connections for the machine presented in Fig. 9
1.1
5.4 2.1
V
4.4 3.1
3.4 4.1
2.4 5.1
1.4 1.2
5.3 2.2
4.3 3.2
3.3 4.2
2.3 1.3 5.2
It is possible to design an equivalent motor using a multilayer concentrated winding

armature by using the same permanent magnet stator, the same commutator and the same
number of brushes (2 or 4 brushes). In fact, one avoids the interleaving of the end-
windings by regrouping simple coils which have the same emf phase on the same tooth of
the armature. This leads to a new armature with only 5 big teeth and 4 simple coils
superposed on the same tooth (Fig. 12). The simple coils can be concentrated around the
tooth to minimize turn length and the copper volume in the end-winding. The connections
of the terminals of the simple coils to the commutator segments and the parallel coil paths
are always identical in both machines (i.e. machines in Fig. 9 and Fig. 12). The width of

the tooth tip influences the electromotive force and can be modified to obtain the same
pattern of the no-load magnetic field in the air-gap.
Fig.11. Diagram of a machine with 20 rotor slots, 4 stator poles, 20 commutator

segments, 2 brushes with a lap winding and a short pitch from 1 to 5
3.1 4.1 5.1 1.2 2.2 3.2 4.2 5.2 1.3 2.3 3.3 4.3 5.3 1.4 2.4 3.4 4.4 5.4 1.1 2.1 3.1
N S N S
N S N S
5.4 1.1 2.1 3.1 4.1 5.1 1.2 2.2 3.2 4.2 5.2 1.3 2.3 3.3 4.3 5.3 1.4 2.4 3.4 4.4 5.4
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Fig.12. Diagram of a multilayer concentrated winding machine with 5 rotor slots, 4 stator
poles, 20 commutator segments
5.1, 5.2 4.1, 4.2 3.1, 3.2 2.1, 2.2 1.1, 1.2
5.3, 5.4 4.3, 4.4 3.3, 3.4 2.3, 2.4 1.3, 1.4
5 4 3 2 1
N S N S
5.4 1.1 2.1 3.1 4.1 5.1 1.2 2.2 3.2 4.2 5.2 1.3 2.3 3.3 4.3 5.3 1.4 2.4 3.4 4.4 5.4
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
It is also possible to make such a concentrated winding armature with only 2 brushes by
using equalized connections on the commutator as shown in the conventional armature of
Fig. 11. In this case, the number of coils can be also reduced to facilitate the
manufacturing process (Fig.13).
Fig. 14 (right) shows a new concentrated winding armature made of a soft magnetic
composite (SMC) material that can replace a 20 slots lap winding armature (Fig 14, left)
made of a stack of laminations in a 180 W/12V motor fan application. The stator, the
brushes and the commutator are the same for both motors. These machines have similar
performance (torque and efficiency) and Table 4 summarizes a comparative analysis on
the material weights. The use of a concentrated winding provides an important weight

reduction of 58 % when compared to the lap winding armature. The armature winding
resistance and the overall armature weight are nearly equal. The magnetic material
weight of the soft magnetic composite armature is increased because the tooth tips are
axially expanded to concentrate the air-gap magnetic flux into the teeth.
Fig.13. Diagram of a machine with 5 rotor slots, 4 stator poles, 20 commutator segments
and 2 brushes with a rotor winding made of concentrated windings wound around the
teeth
5.1, 5.2 4.1, 4.2 3.1, 3.2 2.1, 2.2 1.1, 1.2
5 4 3 2 1
N S N S
5.2 1.1 4.1 5.1 3.2 4.2 2.1 3.1 1.2 2.2 5.2
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
The superior performance of the multi-layer concentrated winding (for current

commutation) has also been confirmed in another study in the case of a 36V supply (Cros
and Viarouge, 2003). The long duration experimental tests at rated operation have shown
that there is no degradation of the commutator and brushes.
Fig 14: 180 W Motor fan with 4 stator poles: lap winding, 20 rotor slots (left);
concentrated winding, 5 rotor slots (right)

Table 4. Comparative characteristics of 180W DC radiator cooling motor (Cros and
Viarouge, 2003)
Armature winding structure Lap winding Concentrated
winding
Number of rotor slots 20 5
Number of stator poles 4 4
Number of brushes 2 2
Commutator segments 20
Brushes 2
Magnetic circuit material Laminations SMC
Weight 422 g 480 g
Copper Weight 130 g 55 g
Total Weight without 552 g 535 g
Shaft and commutator
Synthesis of Efficient Brushed Dc Motor Structures with a Concentrated

Winding
The main requirement for an efficient winding armature is the maximization of the
winding coefficient kb which can be defined as the ratio between the fundamental
component of the magnetic flux embraced by a coil and the total magnetic flux per pole.
This coefficient must be near to unity to maximize the no-load emf amplitude with the
lowest number of turns (cros et al., 2002). Its value is less or equal to unity. Fig. 15
shows the winding coefficient value of several armatures versus the number of slots per
pole.
Fig.15. Variation of the winding coefficient for structures with a concentrated winding
1
0,95
0,9
winding coefficient (kb)
0,85
0,8
0,75
0,7
0,65
0,6
0,55
0,5
0,45
0,6 0,8 1 1,2 1,4 1,6 1,8
Number of slots per pole (s/2p)
One can see that the structures which offer the best winding coefficients are those which
have a number of slots nearly equal to the number of poles. It is always preferable to
choose a high number of stator poles to reduce the mass, however, this number should be
chosen based on a compromise considering the external diameter and the speed. The pole

number has an impact on the number of commutator segments and also on the number of
armature coils. To obtain an efficient brushed DC motor with concentrated winding, we
must select a machine of 2p stator poles and S rotor teeth which respects the following
condition:
S 2 p a with a 1 or 1 or 2 or 3 (1)
With brushed DC motors with concentrated windings, the use a number of commutator
segments (Z) greater than the number of slots is better to minimize current commutation
problems. In this case, there is a plurality of simple coils wound around the same tooth.
The terminals of each coil are connected to different segments for simultaneous
commutation by different brushes. This type of coil arrangement will allow perfectly
balanced parallel paths and avoids any circulating current between these parallel circuits.
The number of segments (Z) is generally equal to the Least Common Multiple (LCM) of
S and 2p:
Z LCM ( S , 2 p ) (2)
Generally, the number of brushes (2B) is equal to the number of stator poles (2p), as
shown by Eq. 3 and N of coils are wound around each tooth (Eq. 4). Consequently, the
number of coils per parallel path Npa is obtained by Eq. 5.
2B 2 p (3)
Z
N (4)
S
Z (5)
N pa
2p
In automotive applications the performance of the motor is sometimes compromised in

order to minimize the manufacturing costs. For example, it may be interesting to reduce
the number of brushes (2B) and the number of coils per tooth (N), particularly in cases of
higher number of stator poles. By employing equalizer connections, this simplification
could be easily realized and the winding would be always well balanced. However, this
modification will affect the life expectancy of the brushes because the current to be
commutated by each brush will increase.
Table 5 presents the main parameters of brushed DC machines for small power
applications which respect the former expressions. The possible simplifications for each
structure are shown in gray-colored cells.

Table 5. Main parameters for DC motor structures with a well balanced concentrated
winding
2P Poles number 2 4 6 6 6 8
S slots 3 5 5 7 8 10
Z commutator segments 6 20 30 42 24 40
N coils per tooth 2 2 4 2 6 2 6 2 6 2 4 8
Npa coils/path 3 5 5 7 4 5
2B brushes 2 2 4 2 6 2 6 2 6 2 4 8
kb winding coefficient 0.87 0.95 0.95 0.97 0.92 0.95
Another possible simplification is to divide the number of commutator segments Z by
two (Eq. 6) and to reduce the number of coils per tooth. This simplification will result in
an unbalanced emf in the different coil paths between brushes but the level of this
unbalance is inversely proportional to the number of coils in each parallel path. The main
parameters of this kind of machine are presented in table 6.
LCM ( S , 2 p )
Z (6)
2
Table 6. Main parameters for downgraded DC motor structures with concentrated

winding
2P Poles 2 4 4 6 6 8 8 8
number
S slots 3 5 5 5 7 7 9 10
Z=LCM(S,2P)/2 3 10 10 15 21 28 36 20
commutator
segments
N coils 1 1 2 1 2 1 3 1 3 1 2 4 1 2 4 1 2 2
per tooth
2B brushes 2 2 4 2 4 2 6 2 6 2 4 8 2 4 8 2 4 8
kb winding 0.87 0.951 0.951 0.951 0.975 0.975 0.984 0.951
coefficient
Synthesis of Efficient Brushless DC Motor Structures with Concentrated

Windings
The selection of a particular combination for the number of poles and the number of slots
in the armature of a brushless dc motor is done in the same manner. First, the number of
rotor poles 2p is chosen based on the operating speed and the rotor external diameter.
This number should be maximised to reduce the mass of the motor, however, it should be
chosen on the basis of a compromise with respect to the electric frequency in order to
limit the magnetic losses. According to the data shown in Fig. 15, the number of slots
should be close to the number of poles to maximize the winding factor and consequently
the motor performance.
The various combinations of slots and poles which allow for the realization of balanced
windings can be determined by Eq. 7 for the case of three-phase armatures. GCD(S,2p) is
the Greatest Common Divisor between the number of slots (S) and the number of poles
(2p) and k is an integer.

S
3k (7)
GCD ( S , 2 p )
The number of slots per pole and per phase is defined by the following equation in the
case of a machine with mph phases.
S
S pp (8)
2 p mph
Table 7 gives a list of the various structures of three-phase machines for which it is
possible to obtain balanced concentrated windings. The number of slots per pole and per
phase (Spp) and the winding coefficient kb are listed in each table cell. The winding
coefficient kb is the performance indicator for each solution. The gray-colored cells
indicate the most efficient structures which present a winding coefficient higher than 0.9.
There is an optimal value of Spp that allows for the maximization of the winding
coefficient kb, for each phase-number. These target values are respectively equal to 1/3
for a 3-phase motor, 1/5 for a 5-phase motor and 1/7 for a 7-phase motor. However, it is
not possible to realize a polyphase structure with a number of slots equal to the number of
poles (S=2p). We can choose however a structure with a number of slots nearly equal to
the number of poles, e.g., S=2p 1 or S=2p 2. These structures have the advantage of
minimizing the cogging torque without slot skewing (Cros et al., 2002). Indeed, the
number of pulsations of the cogging torque for a complete mechanical turn corresponds
to the Least Common Multiple between the number of slots and the number of poles.
Therefore, the frequency of the cogging torque is very high and its amplitude is low.
Table 7. Combinations of number of slots (S) and poles (2p) allowing for the realization
of three-phase machines with balanced concentrated windings
2p 2 4 6 8 10 12 14 16 18 20 22 24
S
3 1/2 1/4 1/8 1/10 1/14 1/16 1/20 1/22
0.866 0.866 0.866 0.866 0.866 0.866 0.866 0.866
6 1 1/2 1/4 1/5 1/7 1/8 1/10 1/11
0.5 0.866 0.866 0.5 0.5 0.866 0.866 0.5
9 3/4 1/2 3/8 3/10 1/4 3/14 3/16 3/20 3/22 1/8
0.617 0.866 0.945 0.945 0.866 0.617 0.328 0.328 0.617 0.866
12 2 1 1/2 2/5 2/7 1/4 1/5 2/11
0.259 0.5 0.866 0.966 0.966 0.866 0.5 0.259
15 5/8 1/2 5/14 5/16 1/4 5/22
0.711 0.866 0.951 0.951 0.866 0.711
18 1 3/4 3/5 1/2 3/7 3/8 3/10 3/11 1/4
0.5 0.617 0.735 0.866 0.902 0.945 0.945 0.902 0.866
21 7/8 7/10 1/2 7/16 7/20 7/22
0.538 0.650 0.866 0.870 0.953 0.953
24 2 1 4/5 4/7 1/2 2/5 8/22
0.259 0.5 0.588 0.760 0.866 0.966 0.958
27 9/8 9/10 3/4 9/14 9/16 1/2 9/20 9/22 3/8
0.423 0.525 0.617 0.695 0.767 0.866 0.878 0.915 0.945
30 5/4 1 5/7 5/8 1/2 3/22
0.389 0.5 0.640 0.711 0.866 0.874

Method for the Determination of the Winding of a Three-Phase Machine
After the selection of the number of slots and poles, the next step is to determine the
winding configuration and the position of the coils in the slots. Generally, for a three-
phase machine, it is preferable to apply a method similar to the one used for the design of
large synchronous machines with a fractional number of slots per pole and per phase.
This method is based on the decomposition of the number of slots per pole and per phase
(Spp). For values below unity, Spp must be reduced to a fraction of two non divisible
integers b and c, as shown in Table 7.
b
Spp (9)
c
A repeatable winding sequence is a list of integer numbers which characterizes the

distribution of the larger and smaller pole-phase coil groups in the armature. This
repeatable sequence can be derived from Spp and Eq. 9. The list has c numbers made of 0
and 1. The number of 1s in the repeatable sequence is equal to b and the number of
0s is equal to b-c. The initial repeatable sequence can then be described as follows:
c

0 0 0 . . . 0 111 . . . 1 (10)

c b b
The configuration of the whole winding structure is derived from 3 consecutive

repeatable sequences under c rotor poles. If c is an even number, the same configuration
can be repeated as a periodic distribution of coils. If c is an odd number, this distribution
is antiperiodic and the direction of the conductors in the coils should be reversed. For a
given structure, one can determine an optimal sequence to find the winding that has the
highest performance (and the highest winding coefficient). This optimal sequence is
derived from the initial one and has the most regular distribution of the number 1s
among the number 0s.
To illustrate this winding determination method, we present an example with a three-

phase machine which has 18 slots and 16 poles. The configuration of the whole winding
can be determined in eight steps as shown in Fig. 16. In the first step, the Spp fraction is
reduced to 3/8 and the initial repeatable sequence of the winding is found to be composed
of five 0s and three 1s. Then, the repeatable sequence is reproduced three times (step
4). In the fifth step, the usual phase sequence AC'BA'CB' is associated to the whole
sequence (A characterizes the return conductor corresponding to conductor A). The
conductors associated to number 1 in the sequence are selected to make the first layer of
winding (step 6). Generally, this layer of winding cannot be directly realized to form a
concentrated winding. In the seventh step, the second layer of the winding is obtained by
reproducing and shifting the initial layer by a tooth or a slot width. The direction of each
conductor must be also reversed. The final configuration of the winding of a three-phase
machine with 18 slots and 16 poles can be checked in the last step. The figure shows only

the winding for 8 rotor poles and this distribution repeats itself for the next 8 poles.
Fig. 16. Determination of the concentrated winding of a three-phase machine with 18

slots and 16 poles (periodic symmetry of the winding under 8 poles)
18 3
Step 1 : S pp
3 14 8
Step 2 : Initial repeatable sequence : 0 0 0 0 0 1 1 1
Step 3 : Optimal repeatable sequence for highest winding performance : 1 0 0 1 0 0 1 0
Step 4: 1 0 0 1 0 0 1 0 1 0 0 1 0 0 1 0 1 0 0 1 0 0 1 0
Step 5: A C B A C B A C B A C B A C B A C B A C B A C B
Step 6 : A A A B B B C C C
Step 7 : A A A B B B C C C
Step 8 :
A A A A A A B B B B B B C C C C C C
N S N S N S N S
ANALYTICAL DESIGN AND FORMULATION OF AN OPTIMIZATION

PROBLEM
An iterative design process for brushed or brushless surface mount PM motors requires a
small computation time for estimation of steady state performances and constraint
evaluations in regard to the application requirements. Generally, the optimization
problem is formulated with a reduced number of variables, an analytical dimensioning
process, and several rapid modeling methods for comparison of different possible
solutions. Fig.17 presents a general flowchart of a design method based on the iterative
optimization process.
The main motor performance requirements impose several non-linear constraints for the
design method. These are the torque (Te), the motor speed (N), the supply voltage (UDC),
the operating temperature (Tw) and the minimal efficiency ( 8.
Other constraints for the maximal motor size are also added by using two parameters: the
maximum external diameter (Dext_max) and the maximum axial length of the motor
(Lmax).
Before beginning the optimization process, one selects the motor structural parameters
i.e. the number of slots (S) and the number of poles (2p), by taking account of the
winding type and the number of phases (mph). Such a decision should also consider the

motor size and the winding performance, represented by the winding coefficient (kb), as
discussed in the preceding sections.
Table 8. Main specification parameters

Torque (N.m) Te
Speed (rpm) N
DC voltage (V) UDC
Maximal external Diameter (m) Dext-max
Maximal motor length (m) Lmax
Working temperature (C) Tw
Minimal efficiency
Different kinds of permanent magnet, soft magnetic and conductive materials can be
selected and characterized by their respective parameters like the magnetic saturation
threshold for the laminations or the residual induction (Br) for the permanent magnets
(Hamdi, 1994). The residual induction and the copper resistivity () are adjusted
according to the motor operating temperature. The maximal flux density B sat in the soft-
magnetic material is fixed to avoid magnetic saturation. Several technological constraints
should also be considered. These parameters impose a limit for certain parameters like
the maximal copper filling factor kr and the minimal air-gap thickness. Generally, the
filling factor does not exceed 0.35 in a small motor and the air-gap thickness is around 1
mm. Table 9 summarizes main material characteristics and technological parameters.
Fig.17. Flowchart of the analytical design method of the motor structure
Motor structure parameters

Number of armature slots: S
Number of permanent magnet poles: 2p
Number of phases: mph
Winding performance coefficient: kb
Non-linear constrained optimization method
Iterative process
Design Analytical design
Variables model
Table 9 Table 10
Specification and
realisation Performance
Constraints Modelling
Tables 7 & 8

A small number of variables including geometrical dimensions, current and magnetic flux
densities are computed by a non-linear constrained optimization procedure that takes
account of the performance criteria and the specifications constraints of the motor
application. These design variables are listed in Table 10. It should be noted that one may
use several similar variables (Bmag) to impose different flux densities in various parts of
the magnetic circuit. Fig. 18 illustrates a number of geometrical stator parameters and
Table 11 shows several equations which can be used to compute the main characteristics
of the motor during the iterative design process.
Table 9. Material parameters and technological constraints

Iron Maximal flux density (T) Bsat Lamination (< 1.8 T) and SMC (< 1.4T)
Residual induction (T) Br Ferrite < 0.4 T and NdFeB bonded (< 0.7 T)
Copper resistivity (.m)
Air-gap thickness (m) e > 0.75 mm
Copper filling factor kr < 0.35
Table 10. Main design variables

Total copper section (m2) Scu
2
Current density (A/m ) J
Armature (airgap) diameter (m) D
Armature length (m) L
Magnet thickness (m) la
Permanent magnet arc ratio 0 1
Iron flux density (T) Bmag Bmag Bsat
Fig.18. Sectional view of a permanent magnet motor with external armature
eca
etb
t De
ecr
e
Dint
Dext
la

Table 11. Analytical design model for a brushless DC motor with concentrated winding
Parameter Analytical expression Constraint relation
Specific armature A J S cu
loading (A/m) D
Air-gap flux Ba Inner Rotor :
B r la
density (T) 2
D
D ln
D 2 ( e la )
External Rotor :
B r la
2
D 2 ( e la )
D ln
D
Electromagnetic T T Te
Torque (Nm) k b s in ( ) 2 D 2 L Ba A
2
Tooth angular width Ba 2

(rad) Bm ag S
Tooth tips thickness etb D Ba 2
(m) t
4 B m ag S
Armature yoke eca D Ba
thickness (m)
4p Bm ag
Permanent magnet ecp D Ba

yoke thickness (m) 4p Bm ag
Slot diameter (m) De External armature :
4 A D 2 etb
D 2 etb
2
S
kr J (1 t )
2
Inner armature :
4 A D 2 e tb
D 2 e tb
2
S
k r J (1 t )
2
External diameter (m) Dext External armature : D e 2 eca Dext Dext max
Inner armature : D 2 ( e la e c p )
Inner diameter (m) Dint External armature : D 2 ( e la e c p ) D int D int max
Inner armature : D e 2 eca
One turn length per Lturn Ba
2L 1 D e D 2 etb
coil for an external S B sa t 4
armature
Number of turns per Ns 30 U AD
p m ph DC
coil
4 2 S T N
Phase Resistance Rph
2 S2 Lturn
Ns2
3 Scu
RMS Phase current Iph 1 Scu J

Ns 2S
RMS No-load flux T
per phase Ns
m ph p I
Total Copper losses Loss
mph Rph Iph2

PERFORMANCE ANALYSIS WITH FINITE ELEMENT AND TIME-
SIMULATION METHODS
The optimal solution must be evaluated by finite element analysis and time-simulation
methods to quantify the analytical model error or to validate the performances before
prototype realization. One can use a finite element model with step by step time
resolution of Maxwell equations. Such modeling method can compute magnetic losses
and takes account of magnetic saturation and rotor motion. When significant differences
on motor performances are observed between finite element and analytical methods,
several correction coefficients for analytical models are derived from these evaluations
and another optimization process is performed. With such a method, the convergence of
optimal analytical design process is achieved and all the constraints imposed by the
application specifications are validated by finite elements or time-simulations (Cros et al.,
2008).
In the case of the finite element calculation of a brushed DC machine, the collector
modeling is based on an equivalent circuit similar to a full bridge converter using a direct
coupling method with meshed armature coils (Fig. 19). Specific switches with an arc
model must be used to reproduce the conduction sequences between brush and collector
segments and to simulate the conduction by the electric arc (Sincero et al., 2010). This
approach is powerful for final solution analysis but it is very time consuming. It is
possible to estimate commutation phenomena more accurately and analyze influence of
some parameters such as the brush angle. To illustrate the performance of such finite
element simulation, the comparison of experimental and simulated armature current
waveforms for a 3 slots-2 poles DC motor with a concentrated winding is presented in
Fig. 20 (Sincero et al., 2010).
Fig.19. Simulation scheme for the finite element simulation of a 3 slots-2 poles DC motor
with a concentrated winding
Another method to model the brush and brushless motor variable speed operation is a
time-simulation method (Fig. 21) based on the matrix resolution of the differential
electrical equations of armature coils (Sincero et al., 2010). The armature winding coils
are modeled by their equivalent circuits composed of self and mutual inductances,

resistances and electromotive forces. These equivalent circuits use constant inductance
values and neglect magnetic saturation. The commutator connects the armature coils to
the input DC source and by Eq. 11, we determine the armature coil current, which is
represented by vector [I]. The electromotive force [E], and the electromagnetic torque T o,
are determined by Eqs. 12 and 13:
d[ I ] d [ ( )]
[V ] [ Rcoil ][ I ] [ L] p (11)
dt d
[ E ] K flux p sin(2 f e p ( )) (12)
[ E ]T [ I ]
T0 Tironloss Tmech (13)

Where [V] is the matrix of coil voltages, [L] is the coil inductance matrix, R coil is the coil
resistance, [] is the armature flux vector, [] is the vector of the spatial phase angles of
the armature coils, p is the pole pair number and is the mechanical angular speed.
Fig.20. Experimental & simulated waveforms for rotor coil voltage, armature current and
DC supply current.
40 Voltage (V) Current (A) 15

30
10
20
10 5
0
-10 0
-20 Rotor coil voltage

-5
DC Current
-30
Rotor Coil current Times (s)
-40 -10
0.025 0.03 0.035 0.04 0.045
The matrix resolution simulation is preferred since it speeds up the time spent on the
performance analysis of PM synchronous motors. Besides, it allows comparing different
structures by only changing the files that generate the parameter matrices. This approach
is more powerful and efficient during a design process where different motors have to be
compared.
The commutator and the control blocks are the systems that differ between a PMBLDC
motor and a PM dc motor in this simulation strategy, although their basic functionalities

are the same. Based on information about the rotor position, the gate signals are
generated to feed the armature coils with a current that is in phase with the emf. This is
necessary to maximize the output torque. However, it is also possible to simulate other
types of control strategies. The simulation method for the commutator of both the
brushless and brushed motors is detailed in (Sincero et al., 2008).
Fig.21. General flowchart of a time-simulation method for brushed and brushless

machines
t
0
dt
The inverter simulation model can neglect the PWM modulation by using the value of the
inverter output voltages averaged over one modulation period. The voltages applied to
the armature coils are positive or negative depending on the rotor position. During the
diode conduction, an amplified current error determines the voltage to be applied in order
to maintain a zero current in the phase (Figueroa et al., 2003). This simulation reproduces
accurately, among others, the waveforms of the machine coils voltages and currents, the
joule and inverter losses, and the torque ripple due to phase commutation effects
(Figueroa et al., 2003).
The average value over each modulation period can be also used to model the dc-dc
converter. This linear variable dc voltage source is connected to the collector model. The
real operation is similar to the inverter one: depending on the rotor position, the positive
or negative input voltage is applied to the armature coils, which is equivalent to the
collector segment contact with the positive or negative brush. The coil under
commutation is short-circuited in order to invert its current direction while maintaining
the output torque constant. If the current commutation is not completed at the end of the
short-circuit interval, an electric arc voltage is generated and applied to the armature coil
which generates an arc current until the complete coil current inversion (Sincero et al.,
2008). The advantage of this methodology is the accurate estimation of the machine
current waveforms, joule and commutation (arc and brush) losses, torque ripple and DC
input current . Fig. 22 shows the brushed dc motor model block diagram.

Fig.22. Brush dc motor model block diagram.
Brush Machine
Input Stage Coupling
Vs
Vb [U ] [V ]
[T u]
Series Collector
Impedance Model
I in Ib [I] [T i ] [J]
Supply
Voltage
[Seq]
Control
C em [J] Currents [J]

calculations
Electro -
Mechanical (Energy
Mechanical
Load storage
Conversion [V ]
[E ] circuits
RL )
Armature Model
Figs. 23 and 24 show the armature coil current and the DC input current for a PMBLDC
motor and PM dc motor, respectively. The steady state waveforms show a good
agreement between experimental and simulation results for both motors.
Fig. 23: PMBLDC motor armature coil current.
15
10
Armature Coil Current (A)
-5
-10
-15
0.005 0.01 0.015 0.02 0.025

Time (s)
Experimental Simulated

Fig. 24: PM DC motor input current.
18
16
14
12
Dc input current (A)

10
4
Simulated
Experimental
2
0
0.031 0.032 0.033 0.034 0.035 0.036 0.037 0.038 0.039
Time (s)
CONCLUSIONS
The constant need to replace mechanical actuated systems and the proposed
move toward a 42V platform, has led to rapid developments in the design of DC
machines for car accessories. The advantages associated with brushless
machines for variable speed applications have also motivated the manufactures
to reconsider the use of traditional brushed DC machines for such applications.
In this context, we have shown that machines with concentrated windings are
particularly efficient. This type of winding is an interesting economical solution
for the minimization of the copper volume or to improve motor efficiency.
The design methodology presented in the preceding sections has been well
validated using several commercial machines. This methodology is general and
all of the steps in the design process for a DC motor (brushed or brushless) have
been described. This process benefits from new calculation techniques
(optimization, modelling) which are increasingly powerful and efficient.
REFERENCES
Cho and Johnston (1999), "Electric motors in vehicle applications", Proceedings of the
IEEE International Vehicle Electronics Conference, pp.193-198 vol.1
Cros and Viarouge (2002), Synthesis of high performance PM motors with concentrated
windings, Energy Conversion, IEEE Transactions on, Volume: 17 Issue: 2 pp. 248 -
253 ISSN: 0885-8969

Cros, Radaorozandry, Figueroa and Viarouge (2008), "Influence of the magnetic model
accuracy on the optimal design of a car alternator", COMPEL, Vol. 27 Iss: 1, 2008,
pp.196 204
Cros, Viarouge, Chalifour and Glinas (2003), "Small brush DC motors using Soft
Magnetic composites for 42V automotive applications", Society of Automotive
Engineers World Congress. SAE03, paper 03M-275
Figueroa, Brocart, Cros and Viarouge (2003), Simplified simulation methods for
polyphase brushless dc motor, Mathematics and Computer in Simulation, vol. 63, p. 209
224
Hamdi (1994), Design of small electrical machines, Wiley & Sons, ISBN 0471952028,
1994
Hultman and Jack (2003), "Soft magnetic composites-materials and applications", paper
presented at the IEEE International Electric Machines and Drives Conference,
IEMDC'03, Vol. 1 pp.516-22
Sincero, Cros and Viarouge (2008), "Arc Models for Simulation of Brush Motor
Commutations, IEEE Trans. on Magnetic, Vol.44, issue 6
Sincero, Ghannou, Cros and Viarouge (2010), Collector model for simulation of brush
machines, Mathematics and Computers in Simulation, vol. 81, issue 2, p. 340 353
Thiemer (2001), "Influence of Automotive 42V Powernet on Small PM DC Motors,"

IEEE Spectrum, pp.591-593
Torrey and Kokernak (2002), "Power Steering: Brushless DC or Switched-Reluctance?",

Power Electronics Technology, pp 24, 33

Development of a CO Sensor for Hydrogen Fuel Cell Powered
Vehicles
G. Neri, G. Leonardi, M. Parthibavarman, V. Hariharan, C. Sekar
ABSTRACT
The development of a CO resistive sensor for application in PEM fuel cells, based on Cu-
doped SnO2 thick film as a sensing layer, has been reported. The synthesis and the
morphological-microstructural characterization of the sensing material were first
described. A resistive sensor was fabricated depositing the nanostructured Cu-SnO2 on a
ceramic substrate provided with interdigited electrodes. The electrical and CO sensing
behavior experiments showed the good performance of the fabricated sensor when
operated in air and nitrogen atmosphere, i.e. high sensitivity in a wide range of CO
concentrations with very fast response/recovery time. The sensor behavior and the
sensing mechanism under simulated reformate fuel conditions (i.e.CO/H 2/H2O mixtures)
were presented and discussed.
INTRODUCTION
Proton-exchange membrane (PEM) fuel cell powered vehicles have the potential to
significantly reduce harmful emissions and dependence on fossil oil fuels [Wee, 2007]. A
hydrogen-based fuel cell could provide the electrical power necessary for a modern
gasoline powered vehicle. However, there are still many obstacles that prevent PEM fuel
cells from playing a major role in electrical power production for transportation purpose.
One of the problems is the supply of hydrogen feed to fuel cell [Dunn, 2002]. The ideal
fuel cell vehicle would operate on pure hydrogen stored on-board. However, storing
hydrogen on-board the vehicle is currently not feasible for technical reasons. Reforming
is an easy way to obtain hydrogen on board the vehicle (Fig. 1), directly from gasoline
[Wipke et al., 2003], or other liquid hydrogen carriers such as methanol.

Figure 1. Schematic view of a vehicle with gasoline or methanol internal reformer for
direct hydrogen production on-board.
But, the presence of carbon monoxide (CO) in the reformate gas, as a by-product of the
reforming process, can poison the PEM Pt-anode by blocking the active sites of the
electro-catalyst and can drastically decrease the output voltage and hence the power
output. CO is contained in the H2-rich reformate exiting the reformer (about 1-3 mole
%), and must be removed or reduced to very low concentrations (less than about 20 ppm)
to avoid poisoning of the anode [Srensen, 2005; Farrell et al., 2007].
Current reformer systems that process hydrocarbon fuels utilize multiple steps in order to
reduce the CO concentration in the hydrogen fuel (Fig. 2).
Fig.2. Schematic view of a gas reformer system for fuel cells.
Hydrocarbon REFORMER SHIFT REACTOR PROX PEM Fuel

fuel 600-900 C ~ 900 C 100-200 C
10-15 % CO 0.2-1 % CO 10-100 ppm CO Cell
The hydrocarbon fuel is first subject to a combination of steam reforming and partial
oxidation that results in a fuel with 15-25% CO content. This fuel then undergoes a water
gas shift reaction that decreases the CO content to approximately 1%. Finally the fuel is
sent through a preferential oxidation (PROX) reactor where the CO content is reduced to
ppm range.
In order to overcome this problem, it is required to continuously monitor the
concentration of CO gas in the reforming gas, by means of CO sensors to be able to
measure CO concentration at various stages of the fuel processor (i.e., for fuel processor
feedback and control) and entering the PEM fuel cell stack (i.e., for fuel cell protection).
Many sensor technologies have been developed for this scope. Most of these sensors are
based on electrochemical devices [Pijolat et al., 2009; Hashimoto, et al., 2002; Planje et
al., 2004; Kirby et al., 2003, Yasuda, 1999]. For example, Pijolat studied a potentiometric

samarium doped ceria (SDC) sensor to detect low CO concentrations in H 2-rich
atmosphere for PEM fuel cell applications [Pijolat et al., 2009]. Solid state devices based
on resistive metal oxide semiconductors (MOS) sensors have also investigated.
Disadvantages of the resistive MOS sensors are the dependence of sensors parameters on
temperature, ambient humidity and their baseline drift with time [Madou and Morrison,
1989]. On the other hand, they are noted for their low threshold of sensitivity, small sizes,
low price and compatibility with state-of-the-art microelectronic technology allowing one
to set on foot large-scale production of such sensors.
However, due to the strong reducing conditions of the reformed gases, the search for a
suitable sensing layer based on metal oxides is a critical aspect. Indeed, despite the wide
variety of MOS sensors available for CO sensing [Williams and Keeling, 1998; Neri et
al., 2008; Kim et al., 2008], only few of them could be adapted for application in strong
reducing ambient. Quick detection capability is another requisite for carbon monoxide
PEM fuel cell sensors. Indeed, the reformer is a highly dynamic system, where the flow
rate and CO content of the H2-rich reformate vary continuously, in response to variations
in the power demands on the fuel cell [Grot et al., 1999]. Actually no chemical sensors
for the detection of CO in reducing gas atmospheres and in particular in H2-rich
atmospheres are commercially available.
NexTech demonstrated a CO sensor based on the change in electrical resistance of copper
(I) chloride thick films [Holt et al., 2002]. SnO2 based sensor have been also investigated.
Tin oxide is the most popular sensing material for CO gas sensors operating in ambient
air [Ihokura and Watson, 1994]. Conventional tin oxide based semiconductor sensors are
not well adapted to work without oxygen and so there exists only a few papers related to
the detection of CO in hydrogen atmosphere [Wurzinger and Reinhardt, 2004; Holt et al.,
2002]. Moreover, SnO2 can be reduced to metallic Sn, demonstrating a poor stability in
reducing atmospheres [Wurzinger and Reinhardt, 2004]. Attempts were made to
overcome this limitation [Prasad et al., 2010; Yamaura et al., 2010; Yamaura et al.,
2011]. The stability and sensing characteristics of SnO2 sensors coated with amorphous
microporous SiBCN layers have been studied in oxygen free, harsh reducing
atmospheres [Prasad et al.; 2010].
Recently, SnO2 doped with Cu and In has been reported to show interesting features for
the detection of CO in hydrogen atmosphere [Prasad et al., 2010; Yamaura et al., 2011].
We have proposed a simple chemical method for synthesizing Cu-doped SnO2
nanoparticles that have demonstrated good performances towards CO in air
[Parthibavarman et al., 2010; Parthibavarman et al., 2011]. In this study, the sensor
performances towards the monitoring of CO in H2-rich environment are reported. It was
found that the presence of hydrogen and humidity put a series of specific challenges and
had a profound effect on the performance of the film. These aspects will be thoroughly
discussed along with hypotheses formulated to explain the sensing mechanisms involved.
EXPERIMENTAL
The morphological, microstructural and electrical characteristics of the Cu-doped SnO2

nanoparticles are performed by using the methods and apparatus described in the
subsequent paragraphs.

Synthesis
A simple chemical precipitation method was employed to prepare the Cu-doped SnO2
nanoparticles. In a typical experimental procedure, the appropriate amount of SnCl4 and
CuSO4 was dissolved in deionized water to make 0.1M solution. Then NHH2O was
added into the solution drop wise under strong stirring until the pH of the solution
reaches to 8. After one hour of continuous stirring, an azury precipitate was obtained.
This precipitate was washed with deionized water several times to remove NH 4+ and Cl-
ions, dried at 105 C for 5 hours, and finally annealed at 600C for 5 hours under ambient
atmosphere. The final powder product (referred as CSN20) was light greenish in color.
Materials Characterization
The prepared Cu.doped SnO2 nanoparticles were subjected to extensive characterization

studies reported in details elsewhere [Parthibavarman et al., 2010]. Powder X-Ray
diffractometry (XRD) analysis was carried out on a Bruker diffractometer within the 2
range of 10 to 80 using CuK as X-ray source ( = 1.5406). The average grain size was
calculated by means of Scherrers equation. Transmission Electron Microscopy (TEM)
and Selected-area Electron Diffraction (SAED) was performed on a Technai G20-stwin
High Resolution Electron Microscope (HRTEM) at an accelerating voltage of 200 kV.
Sensing Tests
A detailed description of the sensor device structure and CO sensing test apparatus has
been reported elsewhere [Neri et al., 2008; Parthibavarman et al., 2011; Krishnakumar et
al., 2009]. Devices for electrical and sensing tests were fabricated by printing films of the
nanoparticles dispersed in water on alumina substrates (6x3 mm2) with Pt interdigitated
electrodes and a Pt heater located on the backside. Pictures of the sensor before and after
the deposition of the sensing material, and the holder device, are reported in Fig. 3.
Figure 3. Pictures of the sensor showing the device (a) before and (b) after the deposition
of the sensing material. The sensor mounted on the holder device is shown in (c).
a) b) c)
The experimental bench for the electrical characterization of sensors in controlled

atmosphere is shown in Fig. 4.

Figure 4. Apparatus for the sensing tests. 1) gas cylinder; 2-3) mass-flow controllers; 4)
bubbler; 5) sensor chamber; 6) sensor; 7) multimeter; 8) power source; 9) computer.
Electrical measurements were carried out by four probe method in the temperature range
from room temperature (RT) to 350C under a dry synthetic air or nitrogen total stream
of 100 sccm. CO gas supplied in certified bottles was further diluted to the required
concentration by mass flow controllers. A multimeter data acquisition unit Agilent
34970A was used for acquiring the sensor signal, while a dual-channel power supplier
instrument Agilent E3632A was employed to bias the built-in heater of the sensor to
perform measurements at high temperatures. The gas response, S, is defined here,
otherwise not specified, as:
S = R/R0 (1)
where R is the electrical resistance of the sensor at different CO concentrations and R 0 is
the baseline resistance. The response time, res, was defined as the time required for the
sensor to reach 90 % of the saturation signal and the recovery time, rec, the time needed
to bring the signal back to 90% of the baseline signal.
RESULTS AND DISCUSSION
Synthesis and Characterization
The synthesis and characterization of Cu-doped SnO2 nanopowders has been reported in a
previous paper [Parthibavarman et al., 2010]. A simple chemical precipitation method, by
using SnCl4, NH3 and CuSO4 as raw materials, has been used for the synthesis. Fig. 5 shows
a TEM image of CSN20 nanoparticles annealed at 600C for 5 hours in ambient
atmosphere.

Figure 5. TEM micrograph of the CSN20 sample and its SAED pattern
The Cu-doped sample consists of nearly spherical shaped nanoparticles with the average
particle size of about 6 nm. SAED pattern confirmed furthermore that the nanopowders are
polycrystalline in nature. Fig. 6 compare the XRD pattern of annealed CSN20 sample, with
that of pure and CSN10 samples.
Figure 6. XRD pattern of the of the (a) pure (b) CSN10 and (c) CSN20 sample.
All the reflections are attributed to tetragonal rutile type SnO2 phase [JCPDS (41-1445)],
with no evidences of other phases. The average crystallite size of CSN20 sample was
estimated to be around 6.5 nm, in good agreement with TEM results, and smaller with
respect to crystallite size of the undoped SnO2 sample (9.5 nm). The results suggest that the
crystallite size reduction is the consequence of the Cu doping [Parthibavarman et al., 2010].
Cu doping was further confirmed by the decrease of the unit cell volume, indicating Sn+4
substitutions by Cu2+ ions, thus creating oxygen-ion vacancies which are charge-
compensated by electron-holes.
Electrical Measurements
Electrical characteristics of the CSN20 film in dry synthetic air, nitrogen, helium and
hydrogen vs. temperature were carried out in the temperature range from RT up to 350 C
(Fig.7).

Figure 7. Electrical characteristics of the CSN20 film vs. temperature in different
atmospheres.
123 C
88 C 134 C 234 C
9
10
Resistance ()
N2
Air
He
H2
8
10
CSN20
50 100 150 200 250 300
Temperature (C)
At room temperature, the resistance of the sensing layer is over the measurement range of
our instrument (above 108 ohm). As the CSN20 film behaves as a semiconductor material,
its resistance decreases with the temperature. It is clearly evidenced that the onset
temperature Tonset, i. e. the temperature at which the baseline resistance begin to be
measurable, strongly depend on the surrounding atmosphere. For such polycrystalline
material, electrical resistance is due to the Schottky barrier height at the grain boundary
which, in turn, is determined by the space-charge layer induced by adsorbed oxygen on the
grain surface. On the basis of this view, the high value of Tonset registered in air (234C), is
related to the high concentration of adsorbed oxygen on the surface, acting as a withdrawing
electron species which increases the barrier height at grain boundaries.
In inert atmosphere, i. e. in nitrogen or helium, Tonset assume much lower values than in air.
Here, the contribution of residual oxygen from nitrogen or helium in the form of impurity
cannot be cannot be considered negligible for the overall conductivity of the sensing film.
On the other hand, in a highly reducing atmosphere, such as in hydrogen, oxygen adsorbed
is instead practically absent due to the reaction with the reducing gas. The shift of Tonset at
low temperature in hydrogen, confirms that adsorbed oxygen governs the film resistance
through the modulation of Schottky barrier height.
CO Sensing Tests
Sensing tests were performed by using the sensor device based on the planar sensing
platform (Fig. 3) with Cu-doped SnO2 sensing layer deposited as a thick film onto platinum
interdigitated electrodes. In practice, by means of the relation (1), the values of resistance
measured provide a measure of the carbon monoxide content in the gas stream. The gas
sensing capability of the Cu-doped SnO2 sample for carbon monoxide was thoroughly
studied by carrying out experiments in synthetic air, nitrogen inert atmosphere and
hydrogen, in both dry and wet environment.

Sensing Tests in Synthetic Air
Fig.8. shows the typical transient response of the CSN20 sensor exposed to 50 ppm of CO
in dry synthetic air at 300C.
Figure 8. a) Typical transient response of the CSN20 sensor exposed to 50 ppm of CO in

dry synthetic air at 300 C; b) calibration curve.
100
a) CSN20 b)
20% O
2
Resistance (M)
Response (R0/R)
10
10
50 ppm CO
T = 300C 1 T = 300 C
1
200 400 600 800 1000 10 100 1000 10000
Time (sec) CO (ppm)
The resistance of the sensing layer decreases after exposure to CO and then the response,
defined as R0/R, is plotted in Fig.8b for various concentration of CO. Data obtained,
compared with that previously reported for the undoped sample [Parthibavarman et al.,
2011], clearly indicated that Cu-loading improve the sensor response, exhibiting high
sensitivity and high signal-to-noise ratio. Further, both the response and recovery times are
lower than those registered on the un-doped one. It is noteworthy that the sensor response
shows a linear dependence on the CO concentration up to the highest concentration value
tested (0.8%, i.e. 8000 ppm).
The response enhancement could be attributed to an increased oxygen deficiency and/or to a
positive catalytic effect due to Cu-doping into SnO2 which accelerates the
adsorption/reaction of CO on the surface. Furthermore, the smaller grain size of the Cu-
doped sensors compared with that of pure SnO2 may also contribute to the improved sensor
response.
Sensing Tests in Nitrogen
The behavior of CSN20 sensor was then tested in inert atmosphere. Fig. 9 shows typical
dynamic responses of the CNS20 sensor exposed to 50 ppm of CO in pure nitrogen at
300C.

Figure 9. Dynamic response of the CSN20 sensor to 50 ppm CO in: a) dry nitrogen; b) wet
nitrogen
10 2,5
a) T = 300C b) T = 300C
50 ppm CO
Resistance (M)
Resistance (K)
N2 2,0
1,5
50 ppm CO N2 + H2O
0,1 1,0
7500 8000 8500 12000 12500 13000
Time (sec) Time (sec)
The tests were carried out either in a dry atmosphere and in the presence of moisture,
saturating the gas with water vapor by means of a bubbler maintained at a constant
temperature of 15 C. The amount of water present in the mixture was about 1.6% by
volume.
Remarkable difference in the sensor response to low carbon monoxide concentration in dry
and wet nitrogen, respectively, was noted. First, we can observe a strong decrease of the
baseline resistance (up to 3 order of magnitude), in the presence of humidity. This is
accompanied by an opposite variation of the resistance baseline when CO is also present. In
addition, the response was significantly reduced than that in dry nitrogen. To explain the
influence of water vapor, it is necessary to have a general understanding of pre-adsorbed
species presents on the oxide surface in dry and wet atmosphere. In dry nitrogen, residual
adsorbed oxygen species (O2-, O-, O2=) are the main species presents on the surface, their
relative abundance depending on the operating temperature. CO reacts with these species
forming carbon dioxide and releasing electron in the bulk, decreasing the resistance (Fig.
10a).
Figure 10. Mechanism of response of the CSN20 sensor to CO in: a) dry nitrogen; b) wet
nitrogen.
a b HCHO
) )
Cu-SnO2 e-
On the other hand, in wet atmosphere, the interaction of water vapor with the semiconductor
surface, results in formation of adsorbed terminal hydroxyl groups, OHads. This explains the

strong decrease of the baseline resistance in the presence of humidity; in fact OHads species
act as electron donors, lowering the baseline resistance [Ihokura and Watson, 1994]. The
reaction mechanism of CO with SnO2 in a humid background is still not fully understood
[Kappler, 2001]. N. Brsan et al [Brsan and Ionescu, 1993] referred that CO can react with
HOads species to give formaldehyde (Fig. 10b). The consequent formation of adsorbed
oxygen, lead to an increase of the resistance [28].
Sensing Tests in Hydrogen
Sensing tests in hydrogen were designed to evaluate the ability of Cu-doped SnO2 sensor to
work in the fuel cell environment. Due to low baseline resistance in hydrogen atmosphere,
sensing tests were carried at lower operating temperature than in air or nitrogen. Then, in
order to operate within the range of SnO2 thermodynamic stability against reduction with an
acceptable sensor response, CO sensing tests in hydrogen were carried out in the range
between 180 and 200 C. Figures 11 compare the response of CSN20 sensor to 5000 ppm
of carbon monoxide in dry and wet hydrogen, respectively.
Figure 11. Dynamic response of the CSN20 sensor to 5000 ppm CO in: a) dry nitrogen; b)
wet nitrogen.
4000 T = 180C 5000 ppm CO / H 2 a)

Resistance ()
3000
2000
1000
CSN20
0
200 400 600 800
Time (sec.)
18
T = 200C H2+ 1,6% H2O b)
Resistance (k)
16
5000 ppm CO 5000 ppm CO
14
12
CSN20
10
0 250 500 750 1000
Time (sec)
The response in wet hydrogen shows an increased signal stability compared to dry
hydrogen, with a rapid response (of the order of 30 seconds or less) and a recovery time

slightly longer. However, the response of the sensor is much lower than that registered in air
or inert atmosphere.
Further tests showed that the behavior of sensing film to CO in dry/wet hydrogen changed
from n- to p-type as a function of temperature and CO concentration. Figure 12 shows the
response and calibration curve obtained for the sample CSN20 operating at 200 C in wet
hydrogen and exposed to pulses of carbon monoxide between 250 and 5000 ppm. It can be
clearly noted that the transition from n-type to p-type semiconductor occurred above a
concentration of about 3500 ppm.
Figure 12. Dynamic response and calibration curve for the sample CSN20 exposed to CO
in wet hydrogen.
20
CSN20
18
5000 ppm CO
Resistance (k) 16 4000 ppm CO
14
3000 ppm CO
12
2000 ppm CO
10 1000 ppm CO
500 ppm CO
T = 200C H2+ 1,6% H2O

8
7000 8000 9000 10000
Time (s)
1,4
CSN20
Sensor response (R/R0)
1,2
1,0
0,8
0,6
T = 200C H2 + 1,6% H2O
0 1000 2000 3000 4000 5000
CO(ppm)
Although still preliminary, the analysis of data obtained with the Cu-doped SnO2 sensor
allows us to draw some general conclusions about the sensing mechanism operating in these
conditions. It was noted that in the presence of water in the feed, the stability of the signal is
improved. The importance of moisture when tin oxide is used as a sensitive material for
detection of carbon monoxide in a strongly reducing atmosphere has already been pointed
out by Yamaura et al. [Yamaura et al., 2010; Yamaura et al., 2011]. Wurzinger and
Reinhardt proposed a mechanism for sensing based on the phenomenon of
adsorption/desorption of water from the surface of the sensitive film [Wurzinger and
Reinhardt, 2004]. This interaction of water molecules with the surface of tin oxide leads to

the formation of surface -OH groups. The removal of hydroxyl groups from the surface of
the material by reaction with carbon monoxide to give rise to carbon dioxide, causes the
electrical resistance increases, as experimentally observed. Experiments proved also that the
consumption of -OH groups is in equilibrium with the generation of the same by
interactions with water [Gromann et al., 2010]
However, the mechanism above presented is not able to explain the phenomenon of
inversion of response to changes in the concentration of carbon monoxide. This particular
phenomenon is due to the change of the concentration of majority carriers in relation to the
interaction of the metal oxide semiconductor with the reducing atmosphere [Gurlo et al.,
2004]. The Cu-SnO2 material in air presents a n-type behavior. Thus, although the presence
of copper induces the formation of holes in the lattice, thereby increasing their concentration
at the expense of electron carriers, the latter remain the majority carriers. In the presence of
strongly reducing atmosphere, i.e. in the presence of small concentrations of O2, the molar
fraction of copper in the lattice of tin oxide, can be however sufficient to cause the n p
transition within the range of CO concentration tested, as we observed. Indeed, when carbon
monoxide interacts with the semiconductor surface, it further removes oxygen atoms from
the surface oxide by increasing the concentration of holes. If the concentration of reducing
species is sufficiently high, the density of holes prevails over the electron carriers. A similar
mechanism has been observed when tin oxide was doped with iron and operates in a
reducing atmosphere for appropriate concentrations of oxygen [Arulsamy et al., 2010].
CONCLUSIONS
Cu-doped SnO2 based CO sensor developed in the present work provides a means to
measure carbon monoxide concentration at the exit of the shift reactor for optimal reformer
operation. The sensing material Cu-doped SnO2 nanoparticles was synthesized by a simple
wet procedure which could be implemented on a large scale. The CO sensor response time
is less than 30 s when it is exposed to step-wise variations of CO.
As the CO sensors operate in harsh humid ambient in practical applications, the sensor
stability is a major concern. The behavior of the newly developed Cu-SnO2 based CO
sensor will be evaluated for a long period during which it will also be subjected to several
shut-down operations.
REFERENCES
Arulsamy AD, Eleri K, Modic M, Cvelbar U, Mozeti M (2010). Reversible carrier-

type transitions in gas-sensing oxides and nanostructures, Chem. Phys. Chem, 11, pp
3704-3712.
Brsan N, Ionescu R (1993). The mechanism of the interaction between CO and an

SnO2 surface: the role of water vapour, Sens. Actuators B 12, pp 71-75.
Dunn S (2002). Hydrogen futures: toward a sustainable energy system, Int. J. of

Hydrogen Energy 27, pp 235264.

Farrell CG, Gardner CL, Ternan M (2007). Experimental and modelling studies of CO
poisoning in PEM fuel cells, J. Power Sources 171, pp 282-93.
Gromann K, Barsan N, Weimar U (2010). Operando DRIFT Measurements on a Tin

Dioxide Based Gas Sensor, Proc. Engineering 5, pp 119122.
Grot SA, Meltser MA, Gutowski S, Neutzler JK, Borup RL, Weisbrod K (1999). Fuel
cell CO Sensor, US Patent 6,001,499.
Gurlo A, Brsan N, Oprea A, Sahm M, Sahm T, Weimar U (2004). An n- to p-type

conductivity transition induced by oxygen adsorption on a-Fe2O3, Appl. Phys. Lett., 85,
pp 2280-2282.
Hashimoto A, Hibino T, Sano M (2002). Solid oxide cells that enable the detection of
CO in reformed gases, Sens. Actuators B 86, pp 1219.
Holt CT, Azad AM, Swartz SL, Rao RR, Dutta PK (2002) Carbon monoxide sensor for
PEM fuel cell systems, Sens. Actuators B 87, pp 414420.
Ihokura K, Watson J. (1994). The Stannic Oxide Gas Sensor, Principles and
Applications, CRC Press, Boca Raton, FL.
Yamaura H, Nakaoka M, Hirao S, Fujiwara A, Yahiro H (2010). CO sensing property of

transition metal oxide-loaded SnO2 in a reducing atmosphere, Mater. Manuf. Process.
25, pp 350353.
Yamaura H, Iwasaki Y, Hirao S, Yahiro H (2011). CuO/SnO2In2O3 sensor for

monitoring CO concentration in a reducing atmosphere, Sens. Actuators B 153, pp 465
467.
Yasuda A, Shimidzu T (1999). Electrochemical carbon monoxide sensor with a

Nafion film, React. Funct. Polym. 41, pp 235-243.
Kappler J (2001). Characterisation of high-performance SnO2 gas sensors for CO

detection by in situ techniques, Ph.D. Thesis, University of Tuebingen, Shaker Verlag.
Kim YS, Hwang IS, Kim SJ, Lee CY, Lee JH (2008). CuO nanowire gas sensors for air
quality control in automotive cabin, Sens. Actuators B 135, pp 298303.
Kirby KW, Chu AC, Fuller KC (2003) Detection of low level carbon monoxide in
hydrogen rich gas streams, Sens. Actuators B 95, pp 224-31.
Krishnakumar T, Jayaprakash R, Pinna N, Donato N, Bonavita A, Micali G, Neri G

(2009). CO gas sensing of ZnO nanostructures synthesized by an assisted microwave
wet chemical route, Sens. Actuators B 143, pp 198-204.

Madou MJ and Morrison SR (1989). Chemical Sensing with Solid State Devices,
Academic Press, San Diego.
Neri G, Bonavita A, Micali G, Rizzo G, Callone E, Carturan G (2008). Resistive CO gas

sensors based on In2O3 and InSnOx nanopowders synthesized via starch-aided solgel
process for automotive applications, Sens. Actuators B 132, pp 224-233.
Parthibavarman M, Hariharan V, Sekar C, Singh VN (2010). Effect of copper on

structural, optical and electrochemical properties of SnO 2 nanoparticles, J.
Optoelectronics Adv. Mater. 12, pp 1894-1898.
Parthibavarman M, Hariharan V, Sekar C, Latino M, Donato N, Neri G (2011). CO gas

sensing behavior of copper-doped tin oxide nanoparticles synthesized by a chemical
precipitation method, Sens. Actuators B (submitted).
Pijolat C, Tournier G, Viricelle JP (2009). Detection of CO in H2-rich gases with a

samarium doped ceria (SDC) sensor for fuel cell applications, Sens. Actuat. B 141, pp 7-
12.
Planje WG, Janssen GJM, de Heer MP (2004). A two-electrode sensor cell for CO
detection in a H2-rich gas, Sens. Actuators B 99, pp 544-555.
Prasad RM, Gurlo A, Riedel R, Hbner MN, Barsan N, Weimar U (2010). Microporous
ceramic coated SnO2 sensors for hydrogen and carbon monoxide sensing in harsh
reducing conditions, Sens. Actuators B 149, pp 105109.
Srensen B (2005). Hydrogen and Fuel Cells: Emerging Technologies and Applications,
Academic Press, New York.
Wee JH (2007). Applications of proton exchange membrane fuel cell systems,

Renewable and Sustainable Energy Reviews 11, pp 17201738.
Williams EW, Keeling AG (1998). Thick film tin oxide sensors for detecting carbon
monoxide at room temperature, J. Mater. Sci. Mater. Electron. 9, pp 51-54.
Wipke KB, Markel A, Haraldsson K, Davis P (2003). Predicting the fuel economy
impact of "cold-start" for reformed gasoline fuel cell vehicles, SAE Transaction 112,
pp. 1795-1804.
Wurzinger O, Reinhardt G (2004). CO-sensing properties of doped SnO2 sensors in H2-

rich gases, Sens. Actuators B 103, pp 104110.
Vohs JM, Barteau MA (1986). Conversion of methanol, formaldehyde and formic acid
on the polar faces of zinc oxide, Surf. Sci. 176, pp 91-114.

The AcademyPublish.org Advantage
The benefits for authors who decide to publish their work with us start at user
friendlywebpage and simplified sign-up form. AcademyPublish takes pride in smooth
publishingprocedure with just a few easy steps, guided throughout by a PPA-publishing
processagent. Open access provides great international impact and worldwide visibility
ofresearch work, while peer reviewhelpsmaintainhigh standards and up-to-dateselectionof
published papers. Academy Publish is also extremely proud of trusted veteran openaccess
staff that makes the difference.
In a nutshell, APs benefits include:
Access free publications to all users (online versions)

High standards of all publications
Expert Editorial Board members and Editors in Chief
Better ROI for public and private institutions
Publishing process guidance
Greater impact and visibility of papers
Easy sing up procedure
International cooperation
Interdisciplinarity
Unlike traditional journals, papers are rarely rejected due to lack of space
Please note that AP publishes only original papers that are not under review by any other
publisher or that have not yet been issued. We strongly believe that is the proper way to
remain competitive thus attracting more views, clicks and downloads that bring authors
citations and distinction to a greater extent.
Registration is easy! Just go to academypublish.org for more information.
Order additional copies of this book at academypublish.org

Questions about our products or your order: contact@academypublish.org
Sign up and publish your manuscript at academypublish.org/site/signup
Contact us:
contact@academypublish.org
AcademyPublish.org
2120 Carey Ave.
Cheyenne, 82001 WY
USA

AcademyPublish.org
Vehicle Engineering
AcademyPublish.org
AP is open to contributions from various and diverse experts. If you are a doctorial
student, an expert in the field, a scholar or an academic and you have great enthusiasm in
publishing papers, you have come to the best open access publisher.
You have come to the right place if you:
-Want to become acknowledged around the world

-Have original paper that you want to publish
-Trust only peer comments and reviews
-Need a strong and trusted publisher name to stand behind your work
-Want worldwide recognition and international audience
WWW.ACADEMYPUBLISH.ORG
ISBN: 978-0-9835850-8-4

Vehicle Engineering

Uploaded by

Document Information

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

Vehicle Engineering

Uploaded by

Copyright:

Available Formats

VEHICLE

Journal of Engineering and Technology,

Open Journal of Economic Research,

It tends to publishpapers covering all branches of conceptual, theoretical and practical

Journal of Computer Sciences and Information Technology,

Journals heterogeneous scope reaches extended topics of theoretical approach to computer

Journal of Medical Research and Science,

AcademyPublish.org is an open access publisher that provides all publication at homepage

Cover Design: AP staff

Dr. Mohsen Mohseni

Automotive Lower Suspension Arm: Experimental and Simulation Durability

Professor Abdullah is the Head of Mechanical and Materials Engineering Department,

Professor Ariffin is an academic staff at the Department of Mechanical and Materials

Mechanical Tolerance Synthesis on the Multimedia Oriented Systems Transport

Michael Vervaeke was born in Asse, Belgium, in 1975. He graduated as Industrial

Youri Meuret graduated in Physics at Ghent University, Belgium, in 1998. In 2004, he

Carl Van Buggenhout was born in Aalst, Belgium, in 1983. He graduated as an

Piet De Pauw. Born in Dendermonde (Belgium) in 1956. He graduated as electronic

Hugo Thienpont was born in Ninove, Belgium, in 1961. He graduated as an electrotechnical

Section 2 Passenger and Transportation Safety and Services

Dr. Susanna Spinsante received her Ph.D. in Electronic Engineering and

Prof. Nouha Taifi is Assistant Professor at the Industrial Engineering Department,

Dr. Mariagrazia Graziano

Section 3 Electrical Engines and Environment Protection

Biolubricants and Triboreactive Materials for Automotive Applications, p.189-p.205

Design Methodology for Small Brush and Brushless DC Motors, p.207-p.235

Development of a Co-Sensor for Hydrogen Fuel Cell Powered Vehicles, p.237-p.250

M. Parthibavarman is currently working as an Assistant Professor in Mahendra Engineering

V. Hariharan is pursuing his doctoral programme in the Department of Physics, Periyar

p. 21 The Current Status of Automotive Coatings and Future Trends Using

p. 53 Automotive Lower Suspension Arm: Experimental and Simulation

p. 75 Mechanical Tolerance Synthesis on the Multimedia Oriented Systems

p. 99 Modeling Window Glass Automotive Antennas with EM Solvers.

p. 113 Spread Spectrum Radar Technology for Automotive Applications: State

p. 149 Monitoring and Controlling Product and Service Engineering: The

p. 163 VHDL-AMS Simulation of an AHRS Platform for Automotive

p. 189 Biolubricants and Triboreactive Materials for Automotive Applications

p. 237 Development of a Co Sensor for Hydrogen Fuel Cell Powered Vehicles

AcademyPublish.org - Vehicle Engineering 17

New Technologies and Advances in Automotive Engineering

The Current Status of Automotive Coatings and Future Trends Using

Automotive Lower Suspension Arm: Experimental and Simulation Durability

Mechanical Tolerance Synthesis on the Multimedia Oriented Systems

Modeling Window Glass Automotive Antennas with EM Solvers. Experimental

AN INTRODUCTION TO AUTOMOTIVE COATINGS

AcademyPublish.org - Vehicle Engineering 21

Figure 1. Specifications of a multilayer automotive system

As Fig. 1 describes, the substrate is firstly coated by a conversion phosphating or

Eco- Friendly Conversion Coatings

22 AcademyPublish.org - Vehicle Engineering

Figure 2: Diagrammatic representation of phosphating reaction (left) and scanning

AcademyPublish.org - Vehicle Engineering 23

Automotive Electrocoat Based Coatings

24 AcademyPublish.org - Vehicle Engineering

Fig 3: Microgels observed by a TEM technique.

Autophoretic conversion coatings (ACC)

AcademyPublish.org - Vehicle Engineering 25

Fig 4: Schematic illustration of the different stages of film formation in an autophoretic

(a) eliminating conversion coating layer like phosphate layer

Eco-friendly Organic Coatings

26 AcademyPublish.org - Vehicle Engineering

AcademyPublish.org - Vehicle Engineering 27

Nanotechnology can be described as the science of molecular engineering. The word

28 AcademyPublish.org - Vehicle Engineering