Lecture Notes in Mechanical Engineering

Károly Jármai
Betti Bolló Editors

Vehicle and
Automotive
Engineering
Proceedings of the JK2016,
Miskolc, Hungary
Lecture Notes in Mechanical Engineering

www.TechnicalBooksPDF.com
About this Series

Lecture Notes in Mechanical Engineering (LNME) publishes the latest develop-

ments in Mechanical Engineering—quickly, informally and with high quality.
Original research reported in proceedings and post-proceedings represents the core
of LNME. Also considered for publication are monographs, contributed volumes
and lecture notes of exceptionally high quality and interest. Volumes published in
LNME embrace all aspects, subfields and new challenges of mechanical
engineering. Topics in the series include:

• Engineering Design
• Machinery and Machine Elements
• Mechanical Structures and Stress Analysis
• Automotive Engineering
• Engine Technology
• Aerospace Technology and Astronautics
• Nanotechnology and Microengineering
• Control, Robotics, Mechatronics
• MEMS
• Theoretical and Applied Mechanics
• Dynamical Systems, Control
• Fluid Mechanics
• Engineering Thermodynamics, Heat and Mass Transfer
• Manufacturing
• Precision Engineering, Instrumentation, Measurement
• Materials Engineering
• Tribology and Surface Technology

More information about this series at http://www.springer.com/series/11236

www.TechnicalBooksPDF.com
Károly Jármai Betti Bolló
•

Editors

Vehicle and Automotive

Engineering
Proceedings of the JK2016, Miskolc, Hungary

123

www.TechnicalBooksPDF.com
Editors
Károly Jármai Betti Bolló
Miskolci Egyetem Miskolci Egyetem
University of Miskolc University of Miskolc
Miskolc, Egyetemvaros Miskolc, Egyetemvaros
Hungary Hungary

ISSN 2195-4356 ISSN 2195-4364 (electronic)

Lecture Notes in Mechanical Engineering
ISBN 978-3-319-51188-7 ISBN 978-3-319-51189-4 (eBook)
DOI 10.1007/978-3-319-51189-4
Library of Congress Control Number: 2017930169

© Springer International Publishing AG 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

www.TechnicalBooksPDF.com
Preface

The production of car and vehicle industry increased greatly in the past decades.
People would like to reach the destination as quickly as possible. The quick
transportation of persons and goods is more and more important. This is the case in
Hungary, where the improvement of the car industry was great in the past decades.
Great car producers settled here like Mercedes Benz, Audi, Suzuki, Opel and also
small and medium enterprises connected to car element production have developed
greatly.
Education has to follow this trend. Vehicle engineering training has a long
tradition in Hungary. At the Budapest Technical University and Economics, at the
István Széchenyi University in Győr they have a long-term experience in this kind
of training. At the University of Miskolc, which is a successor of the Mining and
Metallurgical Academy, the first technical higher educational institution on the
Earth, founded in 1735, the mechanical engineering training started in 1949. The
industrial demand forced the university to start vehicle engineering training also. It
was accredited in 2015 and started this year.
The main requirements for cars and car elements are safety, manufacturability
and economy. Safety against different loads such as permanent and variable actions
is guaranteed by design constraints on stresses, deformations, stability, fatigue,
eigenfrequency, while manufacturability is considered by fabrication constraints.
The economy is achieved by minimization of the cost.
The main topics of the conference are as follows:
Design: Acoustic investigations, Car electronics, Autonomic vehicles, Fatigue,
Industrial applications, Vehicle Powertrain, Modelling and simulation of vehicle
informatics and electronic systems, Vehicle navigation, Visual systems of vehicles,
Mechatronics, Numerical methods FEM and BEM applications, Vibration and
damping, Stability calculations, Structural materials, Structural safety, Structural
connections, Analysis and design of structural elements, Design guides, Fracture
mechanics, Thin walled structures, Driver assist systems, Hybrid and electric cars.
Fabrication: Forming technologies, Surface protection, Production logistics,
Manufacturing technologies, Welding technologies, Heat treatment, Innovative
casting technologies, Industrial applications, Maintenance, Environmental

www.TechnicalBooksPDF.com
vi Preface

protection, Lean technologies, Quality assurance, Gluing technologies, Production,

Testing.
Economy: Life cycle assessment, Fabrication costs, Industrial applications, Cost
engineering, Structural optimization.
Education: Vehicle engineering training, Dual training, Industrial practice,
Training techniques, Training materials.
It is a great pleasure to organize this conference, to give participants an
opportunity to show and discuss the new research results in a friendly atmosphere.
The organizers wish all participants successful days to collect new ideas and get
new acquaintances.

Miskolc, Egyetemvaros, Hungary Károly Jármai

October 2016 Betti Bolló

www.TechnicalBooksPDF.com
Acknowledgements

The editors would like to acknowledge the co-operation and help of the following
organizations

• Hungarian Vehicle Producers Association (MAJOSZ),

• Hungarian Welding Association (MAHEG),
• Hungarian Welding Technology and Material Testing Association (MHtE),
• Foundation for the Development of the Education at the University of Miskolc,
• TÁMOP 4.2.1.B-10/2/KONV-2010-0001 entitled “Increasing the quality of
higher education through the development of research—development and
innovation program at the University of Miskolc supported by the European
Union, co-financed by the European Social Fund.”
• The Mayor and vice mayors of town Miskolc,
• Hungarian Trade and Industrial Chamber in Borsod county (BOKIK),

and last but not least the

University of Miskolc, Hungary, which hosts the conference.
The editors would like to acknowledge the help of the following persons:

László Kota, assistant professor,

Máté Petrik, Ph.D. student,
Éva Engelberth, leader of the dean’s office of the faculty of Mechanical
Engineering and Informatics.

October 2016 Károly Jármai

Betti Bolló

vii

www.TechnicalBooksPDF.com
Contents

Part I Design
Investigation of Rolling Element Bearings Using Time Domain
Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Dániel Tóth, Attila Szilágyi and György Takács
Truck Floor Design for Minimum Mass and Cost Using Different
Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Károly Jármai and József Farkas
Theoretical and Parametric Investigation of an Automobile
Radiator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Máté Petrik, Gábor Szepesi, Károly Jármai and Betti Bolló
Past and Present: Teaching and Research in Vehicle Engines
at the University of Miskolc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Szilárd Szabó, Péter Bencs and Sándor Tollár
Alternating Current Hydraulic Drive the Possibility of Applying
in the Automotive Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Tamás Fekete
Comparative Destructive and Non-Destructive Residual Stress
Measuring Methods for Steering Rack Bar Semi-Product . . . . . . . . . . . . 59
József Majtényi, Viktor Kárpáti, Márton Benke
and Valéria Mertinger
Dynamical Modelling of Vehicle’s Maneuvering . . . . . . . . . . . . . . . . . . . . 69
Ákos Cservenák and Tamás Szabó
Developing a Rotary Internal Combustion Engine Characterised
by High Speed Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
László Dudás

ix

www.TechnicalBooksPDF.com
x Contents

Simulation Methods in the Vehicle Noise, Vibration and Harshness

(NVH) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Károly Jálics
Optimal Damping of Random Excited Systems . . . . . . . . . . . . . . . . . . . . 99
Ferenc Knopp
Application of Knowledge-Based Design in Computer Aided
Product Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
György Hegedűs
Elementary Calculations for Deflection of Circular Rings . . . . . . . . . . . . 115
Géza Németh
Position- and Speed-Dependent, Power-Absorbing Hydraulic
Cylinder with Mathematically Predictable Characteristics . . . . . . . . . . . 123
Lajos Albrecht, Ferenc Mészáros, Szilárd Szabó and Balázs Barna

Part II Technology
Utilization of the GD OES Depth Profiling Technique
in Automotive Parts Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Tamás I. Török and Gábor Lassú
Analysis of Surface Topography of Diamond Burnished
Aluminium Alloy Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Gyula Varga and Viktória Ferencsik
Investigation of Tyre Recycling Possibilities with Cracking
Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Viktória Mikáczó, Andor Zsemberi, Zoltán Siménfalvi
and Árpád Bence Palotás
Utilisation of Various Hydro-Carbon-Based Wastes
by Thermo-catalytic Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
Andor Zsemberi, Zoltán Siménfalvi and Árpád Bence Palotás
Development of Nitrided Selective Wave Soldering Tool
with Enhanced Lifetime for the Automotive Industry . . . . . . . . . . . . . . . 187
Zsolt Sályi, Zsolt Veres, Péter Baumli and Márton Benke
The Effect of Tensile Strength on the Formability Parameters
of Dual Phase Steels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Gábor Béres and Miklós Tisza
Comparison of Two Laser Interferometric Methods
for the Study of Vibrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Miklós Béres and Béla Paripás

www.TechnicalBooksPDF.com
Contents xi

Deburring of Polimer and Metal Matrix Composites . . . . . . . . . . . . . . . . 217

János Líska
The Past, Present and Future of the Training of Internal
Combustion Engines at the Department of Energy Engineering
of BME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Ákos Bereczky
Concept of a New Method for Helical Surface Machining
on Lathe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Dániel Kiss and Tibor Csáki

Part III Electrotechnics, Informatics

Intelligent Transportation Systems to Support Production
Logistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Péter Veres, Tamás Bányai and Béla Illés
A New Scheduling Software for Supporting Automotive
Component Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
Mónika Kulcsárné Forrai and Gyula Kulcsár
The Context Between the Shift of Average Demand
and the Safety Stock of Purchased Parts. . . . . . . . . . . . . . . . . . . . . . . . . . 275
János Korponai, Ágota Bányainé Tóth and Béla Illés
An Overview of Autonomous Intelligent Vehicle Systems . . . . . . . . . . . . 287
Daniel Tokody, Imre János Mezei and György Schuster
Software Reliability of Complex Systems Focus for Intelligent
Vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
György Schuster, Daniel Tokody and Imre János Mezei
Usage of an Optical Flow Sensor in Robotics
to Define Orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
Máté Koba, Roland Bartók and László Czap
Pose Determination for Autonomous Vehicle Control . . . . . . . . . . . . . . . 333
Ahmed Bouzid, József Vásárhelyi, Roland Bartók and László Czap
Description of a Method for the Handling of Customer
Needs in Logistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
Béla Illés, Róbert Skapinyecz and György Wagner
Sensorless Determination of Load Current of an Automotive
Generator Applying Neuro-Fuzzy Methods . . . . . . . . . . . . . . . . . . . . . . . . 355
Csaba Blága

www.TechnicalBooksPDF.com
xii Contents

Distributed CAN-Bus Based Driving Assistance System

on Autonomous Vehicle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
Gergely Kovács and László Czap
Wall and Object Detection with FRI and Bayes-Classifier
for Autonomous Robot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
Roland Bartók, Ahmed Bouzid, József Vásárhelyi and Márton L. Kiss
Optimal Formation of Logistics Networks. . . . . . . . . . . . . . . . . . . . . . . . . 391
György Kovács, Zoltán Varga and Márk Mihalik

Part IV Welding
Development of Complex Spot Welding Technologies
for Automotive DP Steels with FEM Support . . . . . . . . . . . . . . . . . . . . . . 407
László Prém, Zoltán Bézi and András Balogh
A Lightweight Design Approach for Welded Railway
Vehicle Structures of Modern Passenger Coach . . . . . . . . . . . . . . . . . . . . 425
István Borhy and László Kovács
Challenges and Solutions in Resistance Welding of Aluminium
Alloys—Dealing with Non Predictable Conditions . . . . . . . . . . . . . . . . . . 439
Jörg Eggers, Ralf Bothfeld and Thomas Jansen
High Cycle Fatigue Investigations on High Strength Steels
and Their GMA Welded Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
Ádám Dobosy, János Lukács and Marcell Gáspár
Toughness Examination of Physically Simulated S960QL HAZ
by a Special Drilled Specimen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
Marcell Gáspár, András Balogh and János Lukács
Innovation Methods for Residual Stress Determination
for the Automotive Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
Máté Sepsi, Dávid Cseh, Ádám Filep, Márton Benke
and Valéria Mertinger
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499

www.TechnicalBooksPDF.com
About the Editors

Dr. Károly Jármai is Professor at the Faculty of Mechanical Engineering at the

University of Miskolc, where he graduated as a mechanical engineer and received his
doctorate (dr.univ.) in 1979. He teaches design of steel structures, welded structures,
composite structures and optimization in Hungarian and in the English language for
foreign students. His research interest includes structural optimization, mathematical
programming techniques and expert systems. Dr. Jármai wrote his C.Sc. (Ph.D.)
dissertation at the Hungarian Academy of Science in 1988, became a European
Engineer (Eur. Ing. FEANI, Paris) in 1990 and did his habilitation (dr.habil.) at
Miskolc in 1995. Having successfully defended his doctor of technical science thesis
(D.Sc.) in 1995, he subsequently received awards from the Engineering for Peace
Foundation in 1997 and a scholarship as Széchenyi professor between the years 1997–
2000. He is the co-author (with József Farkas) of four books in English Analysis and
Optimum Design of Metal Structures, Economic Design of Metal Structures, Design
and optimization of metal structures, Optimum design of steel structure, and three
monographs in Hungarian, and has published over 610 professional papers, lecture
notes, textbook chapters and conference papers. He has about 770 independent cita-
tions. He is a founding member of ISSMO (International Society for Structural and
Multidisciplinary Optimization), a Hungarian delegate, vice chairman of commission
XV and a subcommission chairman XV-F of IIW (International Institute of Welding).
He has held several leading positions in GTE (Hungarian Scientific Society of
Mechanical Engineers) and has been the president of this society at the University of
Miskolc since 1991. He was a visiting researcher at Chalmers University of
Technology in Sweden in 1991, visiting professor at Osaka University in 1996–1997,
at the National University of Singapore in 1998 and at the University of Pretoria
several times between 2000–2005.

Dr. Betti Bolló is Associate Professor at the Department of Fluid and Heat
Engineering, University of Miskolc, Hungary. She received her M.Sc. degree from
the University of Miskolc in Information Engineering (Systems of Power
Engineering) in 2003. Her research interests include computational fluid dynamics

xiii

www.TechnicalBooksPDF.com
xiv About the Editors

and internal combustion engines. She wrote her dissertation (Ph.D.) at the
Hungarian Academy of Science in 2013. The theme of her dissertation is a
numerical investigation of flow past and heat transfer from a heated circular
cylinder.

www.TechnicalBooksPDF.com
 Part I
Design

www.TechnicalBooksPDF.com
Investigation of Rolling Element Bearings
Using Time Domain Features

Dániel Tóth, Attila Szilágyi and György Takács

Abstract Rolling element bearings can be found widely in domestic and industrial
applications. They are important components of most machinery and their working
conditions influence the operation of the entire machinery directly. Bearing failures
may cause machine breakdown and might even lead to catastrophic failure or even
human injuries. In order to prevent unexpected events, bearing failures should be
detected as early as possible. Different methods are used for the detection and
diagnosis of bearing defects. These techniques can be classified as noise analysis,
acoustic measurements, wear debris detection, temperature monitoring, vibration
analysis etc. Vibration signals collected from bearings carry detailed information on
machine health conditions. This paper deals with a bearing test procedure which
based on vibration analysis.

1 Introduction

Vibration monitoring is one of the essential tool that allows to determine the
mechanical health of different components in a machine. When the assessment of a
ball bearing is performed by vibration analysis, several signal processing techniques
can be considered. These techniques can be performed within either the time or the
frequency ranges. Among these methods the time domain features are the most
appropriate with random signals, where other signal analysis methods are not
suitable. These methods facilitate fast data processing and computation. Numerous
time domain statistical parameters have been used as trend parameters to detect the

D. Tóth (&)
A. Szilágyi
G. Takács

University of Miskolc, Miskolc, Hungary
e-mail: toth.daniel@uni-miskolc.hu
A. Szilágyi
e-mail: szilagyi.attila@uni-miskolc.hu
G. Takács
e-mail: takacs.gyorgy@uni-miskolc.hu

© Springer International Publishing AG 2017 3

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_1

www.TechnicalBooksPDF.com
4 D. Tóth et al.

bearing failures. The most frequently applied stochastic features are the
root-mean-square (RMS) value, peak value, skewness, impulse factor, shape factor,
clearance factor, crest factor and kurtosis [1, 2].

2 Bearing Test Device

Rolling element bearing condition monitoring can be accomplished by using a test

instrument. Such a device is located at University of Miskolc, Department of
Machine Tools. The test device is illustrated in Fig. 1.
The equipment is suitable for performing the bearing fatigue and measurement
investigations. The particular symbols have the following meanings:

Fig. 1 Experimental test rig

www.TechnicalBooksPDF.com
Investigation of Rolling Element Bearings … 5

• 1: three-phase motor,
• 2: rigid table,
• 3F: supporting bearings of fatigue side,
• 3M: special supporting plain bearings of measurement side,
• 4F: fatigued bearing position,
• 4M: measured bearing,
• 5: double-acting hydraulic cylinder,
• 6: load cell, the adjustment of hydraulic load,
• 7F: fatigue test shaft,
• 7M: measurement test shaft,
• 8: length ribbed belt,
• 9: belt tensioner,
• 10: piezoelectric vibration accelerometer.
During the measurements the “7M” shaft works at the given rotational speed
(1500 min−1), while the “6” hydraulic cylinder exerts artificial load (1 kN) for the
“4M” bearing.

3 Description of Investigation

Fundamentally, two proceedings are used for the experimental analysis of rolling
element bearings. One method is the fatigue test when the bearings operate until they
get permanent damage, and we measure their vibration trends meanwhile. However,
the process takes relatively long time, but it can be accelerated with the bearing
overload and increased rotational speed. Another technique is the production of one or
more artificial failure of the elements of bearings. In this case the vibration signal
should be measured and compared to data of faultless bearings. According to the
literature [3–5], generally this may use methods such as spark erosion, acid etching,
scratching or mechanical indentation. In this research, we used a well reproducible
method to create artificial faults. A Rockwell hardness tester applied to make defects
to the inner ring of bearings. This method needs a bearing with plastic cage, because it
should be disassemble and assemble non-destructively. Figure 2 shows the ball
bearing type 6303 which used during experiments.

Fig. 2 The test bearing and

the artificial defect on the
inner ring

www.TechnicalBooksPDF.com
6 D. Tóth et al.

As it was written previously a Rockwell hardness tester is used to cause local

defects. The type of this machine is HR—150A. It is suitable for examining the
effects of three types of loads. The major load values are 60, 100 and 150 kg. The
effects of each loading were examined more than 10 times.
Optical examination can be applied to measure failure size. Polarising micro-
scope is widely used for higher resolution. Carl Zeiss Jenavert polarising micro-
scope is applied to inspect the extent of the defect.
The average extent of the failure is 265 lm in diameter in case of 60 kg load,
411 lm in diameter in case of 100 kg load and 478 lm in diameter in case of
150 kg load. The following illustrations show the effects of different loads (Fig. 3).

Fig. 3 Inner ring defects in case of 60 kg, 100 kg and 150 kg load (15 times magnification)

www.TechnicalBooksPDF.com
Investigation of Rolling Element Bearings … 7

4 Analysis of Measurements

During the experiment, first of all the vibration patterns were measured from the
examined bearing using piezoelectric vibration accelerometer (the type of it is
Kistler 8632C50). After that the artificial defect was created and vibration patterns
were measured again. It is followed by time-domain tests during which statistical
features have been calculated. These stochastic indexes can be calculated by using
the formulas below (Fig. 4).
The measurement cycles are performed at 9.6 kHz sampling frequency. Five
vibration samples and 16,384-element samples were taken within each cycle.
Statistical features were calculated based on sampled values. These parameters were
computed by a program code, which runs in Maple mathematical software. Table 1
contains the statistical parameters in case of 60 kg load.
Table 2 includes the stochastic features in case of 100 kg load.
It is visible that the most of the parameters have doubled under this load. Table 3
contains the statistical parameters in instance of 150 kg load.
It is clearly visible that the statistical parameters of a defective bearing tend to be
higher than the values of a normal bearing. The percentage increase is depicted in
Fig. 5.
According to the graph it is clear that the Standard deviation, the Peak value and
the RMS were the most sensitive to this artificial error. Nevertheless, it is obvious
that the Kurtosis and the Skewness also have good correlation.

Fig. 4 Calculation of
stochastic features [1]

www.TechnicalBooksPDF.com
 8

Table 1 Statistical parameters in new and damaged conditions (60 kg load)

Peak value RMS Kurtosis Skewness Standard deviation Peak value RMS Kurtosis Skewness Standard deviation
1 0.5688 0.0761 3.9468 1.0173 0.0461 1 0.6569 0.0792 4.3318 1.0370 0.0480
2 0.6150 0.0813 3.8500 1.1566 0.0485 2 0.6743 0.0831 4.1450 1.1802 0.0491
3 0.5632 0.0784 3.9200 0.9854 0.0463 3 0.6242 0.0790 4.4052 1.0209 0.0475
4 0.6369 0.0759 4.2139 1.0142 0.0449 4 0.6840 0.0784 4.7605 1.0403 0.0467
5 0.6807 0.0730 4.3995 1.0921 0.0442 5 0.7643 0.0802 4.8643 1.3295 0.0480
6 0.5376 0.0746 3.9534 1.0352 0.0523 6 0.6421 0.0755 4.5431 1.0723 0.0542
7 0.6712 0.0865 3.7533 0.8960 0.0508 7 0.7823 0.0921 4.0854 0.9302 0.0546
8 0.5114 0.0864 3.9900 1.0142 0.0441 8 0.6432 0.0887 4.3586 1.0274 0.0452
9 0.5255 0.0976 3.9631 1.0532 0.0553 9 0.6422 0.0992 4.2961 1.0689 0.0568

www.TechnicalBooksPDF.com
10 0.5931 0.0867 3.4861 0.8991 0.0521 10 0.7053 0.0888 3.8716 0.9582 0.0529
11 0.6274 0.0974 3.8255 1.0809 0.0580 11 0.7255 0.1032 4.2016 1.0904 0.0583
D. Tóth et al.
 Table 2 Statistical features in new and damaged conditions (100 kg load)
Peak value RMS Kurtosis Skewness Standard deviation Peak value RMS Kurtosis Skewness Standard deviation
1 0.6225 0.0827 3.9236 1.1359 0.0477 1 1.1337 0.1482 5.2670 1.6243 0.0906
2 0.6508 0.0776 4.6827 1.1124 0.0471 2 1.1450 0.1425 6.1030 1.4740 0.0884
3 0.7425 0.0806 4.9360 1.1823 0.0495 3 1.3751 0.1461 6.7832 1.7498 0.0920
4 0.6674 0.0701 4.7007 1.3366 0.0435 4 1.1821 0.1260 6.5078 1.8236 0.0778
Investigation of Rolling Element Bearings …

5 0.6150 0.0813 3.8500 1.2566 0.0485 5 1.0752 0.1487 4.7519 1.6767 0.0861
6 0.6569 0.0779 4.3318 1.0370 0.0463 6 1.1550 0.1385 5.2563 1.3732 0.0820
7 0.5618 0.0628 4.8172 1.2033 0.0385 7 1.0103 0.1120 6.2307 1.6778 0.0728
8 0.5970 0.0819 3.8334 0.9852 0.0488 8 1.0275 0.1456 4.9583 1.3146 0.0873
9 0.6475 0.0757 4.3553 1.0751 0.0459 9 1.1539 0.1358 5.5933 1.4559 0.0848

www.TechnicalBooksPDF.com
10 0.6527 0.0763 4.4994 1.0673 0.0455 10 1.1633 0.1360 5.6846 1.4334 0.0832
11 0.6705 0.0782 4.5762 1.2543 0.0484 11 1.1949 0.1366 5.8274 1.7112 0.0915
9
 10

Table 3 Statistical indexes in new and damaged conditions (150 kg load)

Peak value RMS Kurtosis Skipness Standard deviation Peak value RMS Kurtosis Skewness Standard deviation
1 0.5824 0.0681 4.0148 0.9524 0.0406 1 1.5039 0.1683 6.0793 1.5524 0.1133
2 0.6386 0.0747 4.2957 1.0603 0.0453 2 1.4868 0.1816 6.5047 1.6997 0.1244
3 0.6777 0.0847 4.3517 1.1184 0.0554 3 1.5195 0.2044 6.5895 1.8040 0.1561
4 0.5527 0.0644 3.7721 1.0339 0.0428 4 1.4658 0.1561 6.0891 1.6297 0.1203
5 0.6292 0.0703 4.9252 1.3477 0.0431 5 1.5429 0.1570 7.3456 2.1513 0.1215
6 0.5875 0.0696 3.8739 0.9273 0.0415 6 1.4876 0.1402 6.1514 1.4710 0.1135
7 0.7160 0.0914 4.4818 1.4629 0.0564 7 1.5337 0.1945 6.5947 2.2913 0.1563
8 0.6309 0.0663 4.4436 1.2635 0.0411 8. 1.4771 0.1474 6.8939 2.0170 0.1122
9 0.6043 0.0839 4.4436 1.1636 0.0588 9 1.3790 0.1957 7.0717 1.8737 0.1592

www.TechnicalBooksPDF.com
10 0.5925 0.0825 4.1238 1.0366 0.0487 10 1.3913 0.1752 6.8102 1.6496 0.1334
11 0.6391 0.0855 4.4342 1.1429 0.0518 11 1.4432 0.1944 7.4114 1.8093 0.1425
D. Tóth et al.
Investigation of Rolling Element Bearings … 11

Fig. 5 The percentage increase of loads

5 Conclusion

Trustworthy and accurate measuring methods and devices are inevitable for rotary
and bearing condition monitoring. The investigation of vibration signals is a sig-
nificant technique for monitoring the condition of machine components. Stochastic
parameters are widely used as features in failure diagnostics. Present paper shows
that the time domain techniques can be effectively used in condition monitoring and
fault diagnosis of ball bearings. These methods are reliable tools and they make
possible fast data processing.

Acknowledgements This research was supported by the ÚNKP-16-3 New National Excellence
Program of the Ministry of Human Capacities.

References

1. Patel J, Patel V, Patel A (2013) Fault diagnostics of rolling bearing based on improve time and
frequency domain features using artificial neural networks. IJSRD 1(4)
2. Patidar S, Soni PK (2013) An overview on vibration analysis techniques for the diagnosis of
rolling element bearing faults. IJETT 2013

www.TechnicalBooksPDF.com
12 D. Tóth et al.

3. Kharche PP, Kshirsagar SV (2014) Review of fault detection in rolling element bearing.
IJIRAE 1(5)
4. Patkó Gy, Takács Gy, Demeter P, Barna B, Hegedűs Gy, Barak A, Simon G, Szilágyi A (2010)
A process for establishing the remanent lifetime of rolling element bearings. In: XXIV
microCAD International Scientific Conference, Miskolc (Hungary), March 2010
5. Howard I (1994) A review of rolling element bearing vibration detection, diagnosis and
prognosis. DSTO-RR-0013

www.TechnicalBooksPDF.com
Truck Floor Design for Minimum Mass
and Cost Using Different Materials

Károly Jármai and József Farkas

Abstract In the chapter the floor structure of a truck produced by a company in

Hungary has been investigated. The structure consists of steel members, or extruded
Al-alloy longitudinal and cross members as well as a tread deck plate. Using an
optimum design process, namely the Hillclimb optimizer, significant mass and cost
savings may be achieved by decreasing the deck plate thickness and changing the
profile, dimensions and number of cross members. Comparison is made using the
combination of the steel and aluminium, or using only steel alone. Design con-
straints relate to fatigue stress range of welded joints, to local buckling of extruded
or normal profiles and to fabrication size limitations. A special loading case is also
considered when a wheel is staying on a curb and the floor is distorted.

1 Introduction

There are some trucks for beverage transport, where the truck structure has a steel
chassis consisting of two longitudinal beams. The subframe is constructed from two
longitudinal beams bolted on steel beams. They can be made from Al-alloys, or
structural steel. The Al-alloy floor structure has three layers as follows (Fig. 1):
cross members welded to subframe, the longitudinal members welded to cross
members, tread deck plate distributing the pallet loads. The material of cross
members is an Al-alloy AlMgSi0.7 according to German standard DIN 1725 [1] of
Rp.0.2 = 215 MPa according to DIN 1748 [2] (international alloy type 6005A). The
tread deck plate material is an Al-alloy AlMg2.5 (international alloy type 5052).
These main structural parts are framed by side rails, which carry the loads from

K. Jármai (&)
J. Farkas (Deceased)

University of Miskolc, Miskolc, Hungary
e-mail: jarmai@uni-miskolc.hu
J. Farkas
e-mail: altfar@uni-miskolc.hu

© Springer International Publishing AG 2017 13

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_2

www.TechnicalBooksPDF.com
14 K. Jármai and J. Farkas

Fig. 1 Truck floor structure

roof, sidewalls and doors. We have made an optimization using aluminium, or

normal steel in the floor structure. Due to the fact that the fatigue limit for the steel
at Eurocode 3 up to 690 MPa and at IIW recommendation up to 960 MPa does not
change, it does not worth to use higher strength steels, only normal structural steel.
Our aim is to decrease the material cost of the floor structure by changing the
profile, dimensions and number of cross members, the thickness of deck plate as
well as the material grades.

2 Load Cases

2.1 Loads in the Horizontal Floor Position

Two load cases should be considered in the design of cross members as follows:
(a) loads due to pallets, roof, door and side walls in the horizontal floor position;
(b) the same loading as in (a) but a wheel is staying on a curb, thus, the floor is
distorted.

www.TechnicalBooksPDF.com
Truck Floor Design for Minimum Mass … 15

Fig. 2 Loads on the

cantilever part of cross
members

Loads acting on an outside cross member are as follows:

a corner column 205 N

roof 2060/4 515 N
upper door 1420/2 710 N
front wall 1033/2 516 N
F1 = 1946 N

Load from pallets: mass of a pallet is Fp = 8500 N, intensity of the uniformly

distributed load is p = Fpnp/(BL), where the number of pallets placed on the half
floor np = 5, B and L are the dimensions of a half cantilever floor surface. The
uniformly distributed normal load acting on a cross member is pc = pL/(nc−1), nc is
the number of cross members.
The maximum bending moment in a cross member is (Fig. 2)

pc B 2 F p np B
Mmax ¼ þ F1 B ¼ þ F1 B ð1Þ
2 2 ð nc  1 Þ

Calculating with Fp = 8500 N, np = 5, B = 720 mm, F1 = 1946 N one obtains

bending moments for different numbers of cross members. This number is limited
by the dimension of pallets (800 mm) to nc.min = 10. Since the original number of
cross members is 14, we calculate with nc = 14, 12 and 10. For these values of nc
one obtains

M14 ¼ 2:578; M12 ¼ 2:792 and M10 ¼ 3:1011 kNm:

The corresponding shear forces are as follows:

Q ¼ Fp np =ðnc  1Þ þ F1 ; Q14 ¼ 5215; Q12 ¼ 5810 and Q10 ¼ 6668 N:

2.2 Loads on the Distorted Floor

Measurements have been carried out on a truck loaded with pallets and with a
wheel staying on a curb in a height of 91 mm. The measured deflections have

www.TechnicalBooksPDF.com
16 K. Jármai and J. Farkas

Fig. 3 Measured deflections of a distorted cross member, when a left truck wheel is staying on a
curb

shown that the cross members near the wheel being lifted up are loaded by bending
as it is seen on Fig. 3. This cross member can be modelled as a cantilever beam of
its whole length Lc loaded by a force F corresponding to a deflection w. This
deflection can be approximately calculated as w ¼ 138  Lc u, where
Lc = 2427 mm, uðradÞ ¼ 2:91 p=180 ¼ 0:0508, thus, w = 15 mm. Furthermore

3EIx w
F¼ ; Mc:max ¼ FLc ð2Þ
L3c

where E = 7 x 104 MPa is the elastic modulus of aluminium, E = 2.1 x 105 MPa
for steel, Ix is the second moment of area.

3 Geometric Characteristics of Cross Members

The cross-section loaded by bending and shear consists of a cross member and a
part of the deck plate (Fig. 4). We calculate an effective width of the deck plate 50t,
t is the thickness. In the case of a rectangular hollow section (RHS) the geometric
characteristics of this cross section are as follows [3]:

A ¼ A1 þ A2 ; A1 ¼ 2htw þ 2btf ; A2 ¼ 50t2 ð3Þ


A1 h þ t t
yG ¼ þc ; yc ¼ h þ c þ  yG ð4Þ
A 2 2


h3 tw btf h2 h 2
Ix ¼ þ þ A 1 yc  þ A2 y2G ð5Þ
6 2 2

www.TechnicalBooksPDF.com
Truck Floor Design for Minimum Mass … 17

Fig. 4 Cross-sections of cross members

In the case of I-profile (Fig. 4) the characteristics are as follows:

A1 ¼ htw þ 2btf ð6Þ


h3 tw btf h2 h 2
Ix ¼ þ þ A 1 yc  þ A2 y2G ð7Þ
12 2 2

In our previous calculations [4] we have made comparisons using the rectangular
hollow section, I- and C-profiles. It was found that the best cross section is the
I-beam. That is the reason why the I-profile has been chosen.

4 Design Constraints

4.1 Constraints on Fatigue Stress Range for the Horizontal

Floor Position
Mmax DrN
r1 ¼ ymax  ; ymax ¼ maxðyG ; yc Þ ð8Þ
Ix cMf

Q DsN
s1 ¼  ; ð9Þ
Aw cMf

where Aw = htw for I-profile.

Since the cross members are welded to longitudinal subframe beams, they should
be designed considering the fatigue of welded joints. According to Hobbache [5] the
fatigue stress range for number of cycles 2 x 106 in the case of transverse stiffener

www.TechnicalBooksPDF.com
18 K. Jármai and J. Farkas

welded on girder web (detail 512 for structural aluminium alloys) is DrC ¼ 28 MPa.
Calculating with a realistic number of cycles N = 2 x 105,

1 2x106
log DrN ¼ log þ log DrC ¼ 1:78049; DrN ¼ 60:3 MPa ð10Þ
3 2x105

For steel DrC ¼ 80 MPa (detail 512 for structural steel, the same as for Al)
DrN ¼ 172:3 MPa. With a safety factor of 1.25.
For aluminium

DrN 60:3
¼ ¼ 48:2 MPa ð11Þ
cMf 1:25

For shear it is

DsN 44:3
DsC ¼ 28; DsN ¼ 44:3; ¼ ¼ 35:44 MPa ð12Þ
cMf 1:25

For steel

DrN 172:3
¼ ¼ 137:8 MPa ð13Þ
cMf 1:25

For shear it is

DsN 126:8
DsC ¼ 80; DsN ¼ 126:8; ¼ ¼ 101:44 MPa ð14Þ
cMf 1:25

It should be mentioned that we calculate with the bending moment also from
static load F1 in the fatigue constraint as an approximation in the safe side.

4.2 Constraint on Fatigue Stress Range for the Distorted

Floor Position
Mc:max 3Ew DrN1
r2 ¼ ymax ¼ 2 ymax  ð15Þ
Ix L cMf

In the case of distorted floor position the maximum bending moment arises at the
end of the cross member, where it is welded to subframe by fillet welds. For this
joint, according to [5] (detail No.413) DrC1 ¼ 22 MPa and a realistic number of
cycles N = 105 it is

www.TechnicalBooksPDF.com
Truck Floor Design for Minimum Mass … 19

DrN1 59:7
¼ ¼ 47:7 MPa ð16Þ
cMf 1:25

4.3 Constraints on Local Buckling of Profiles

Web of I-section (unreinforced)

For aluminium

h=tw  22e=g; ð17Þ

For steel [6]

h=tw  69e=g; ð18Þ

g ¼ 0:65 þ 0:35 yy0c when 1 y0

yc 0
g ¼ 0:65 þ 0:30 yy0c when 0 y0
yc  1

t
y0 ¼ yG   c ð19Þ
2
t
yc ¼ h þ c þ  yG ð20Þ
2

Flange of I-section (unreinforced)

For aluminium alloy

b=tf  14e ð21Þ

For steel

b=tf  28e ð22Þ

For aluminium
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
250 235
e¼ ; for steel e ¼ ð23Þ
rmax =cM1 rmax =cM1

4.4 Fabrication Constraints: Size Limitations

Some constant dimensions are prescribed by the original structure as follows:

www.TechnicalBooksPDF.com
20 K. Jármai and J. Farkas

h ¼ 100; c ¼ 34 mm ð24Þ

The web thickness is limited to

tw:min ¼ 3:4 mm ð25Þ

to guarantee the quality of welding.

The tread plate thickness is limited to

tmin ¼ 2 mm ð26Þ

Since the cross members should be welded to side rails, the extruded shapes
should not have any reinforcing ribs or bulbs, since they are in the way of welding.
It should be mentioned that the extruded I-profiles with or without reinforcing
ribs or bulbs optimized for pure bending have the same minimum cross-section
area, thus, the use of ribs or bulbs does not result in mass savings.

5 Optimization Characteristics and Results

The objective function is the cross-sectional area of the cross members and deck
plate part (Eq. 3).
The unknown variables are the dimensions of profile flanges b and tf.
The constraints are as follows: Eqs. 11, 12, 13, 14, 15, 21, 22, 24, 25, 26.
The optimization is performed for I-profile and for three numbers of cross
members nc = 14, 12 and 10.
Mathematical method: the Rosenbrock’s Hillclimb algorithm is used [7].
Results are summarized in Table 1 and 2.

Table 1 Optimum sizes of the frame

Original aluminium Optimized Optimized
structure aluminium steel
Structural weight in 304.48 172.31 239.7
kg
Structural cost in 927.44 550.27 247.40
USD

www.TechnicalBooksPDF.com
Truck Floor Design for Minimum Mass … 21

Table 2 Result of the Profile nc = 14 nc = 12 nc = 10

optimization
Aluminium b 55 60 65
I-profile tf 7.2 7.2 7.8
A1 1332 1404 1536
mc kg 104.41 95.18 89.20
KT $ 927 927 927
kT $ 0.54 0.63 0.795
Kc $ 304.74 291.78 260.64
Steel b 30 30 35
I-profile tf 2 2 2.1
A1 660 660 747
mc kg 123.35 105.73 93.28
KT $ 187 187 187
kT $ 0.109 0.128 0.153
Kc $ 269.77 252.15 247.4

6 Mass Savings

The mass of the original tread plate of thickness t = 4.5 mm and dimensions
2280  6750 mm, taking the density of Al-alloy q ¼ 2:7  106 kg/mm3, for
steel q ¼ 7:85  106 kg/mm3

mpl ¼ qtð2280  6750Þ ¼ 186:98 kg:

For aluminium
Mass of the optimized Al plate of t = 2.0 mm is mpl.opt = 83.11 kg.
The mass of Al cross members can be calculated as

mc ¼ qA1 nc Lcm

where Lcm = 2440 mm is the length of a cross member. The calculated masses are
shown in Table 2.
The original mass of the tread plate and cross members is

m ¼ mpl þ mc ¼ 186:98 þ 117:50 ¼ 304:48 kg:

The mass of the optimum Al solution is mmin = 83.11 + 89.20 = 172.31 kg, the
mass saving is 132.17 kg for one truck (43%).
For the steel
Mass of the optimized steel plate of t = 2.0 mm is mpl.opt = 146.424 kg.

www.TechnicalBooksPDF.com
22 K. Jármai and J. Farkas

The mass of steel cross members can be calculated as

mc ¼ qA1 nc Lcm

The mass of the steel tread plate and cross members is

m ¼ mpl þ mc ¼ 146:42 þ 93:28 ¼ 239:7 kg:

7 Cost Savings

For aluminium
Cost of tread deck plate

London Metal Exchange (LME) price of aluminium [8] 1.559 $/kg

surcharge 0.9568 $/kg
total 2.5159 $/kg

Cost of cross members

Cost of the original plate (t = 4.5) 186.98  2.5159 470.44 $

Cost of the optimized plate (t = 2 mm) 83.11  2.5159 209.09 $

Cost of cross members

LME aluminium total 2.5159 $/kg

extrusion work upcharge 1.3004 $/kg
total kc = 3.8163 $/kg

The total cost, including the proportional tool cost can be expressed as

KT ¼ kc mc þ kT

where

KT
kT ¼
50nc Lcm

www.TechnicalBooksPDF.com
Truck Floor Design for Minimum Mass … 23

KT is the tool cost, 50ncLcm is the total length of extruded bars for 50 trucks (one
year production).
The results of the calculations are shown in Table 2.
The total cost of the original deck plate and cross members is
K = 470.44 + 457.00 = 927.44 $
and that of the optimum (10 cross members of I-profile) Kmin = 209.09 +
341.18 = 550.27 $
The cost savings for one Al truck is 377.17 $ (39%)
For steel
Cost of tread deck plate

London Metal Exchange (LME) price of steel 0.375 $/kg

surcharge 0.231 $/kg
total 0.606 $/kg

Cost of the optimized steel plate (t = 2 mm) 241.623  0.606 146.424 $

Cost of cross members

LME steel 0.606 $/kg

cutting and welding costs 0.475 $/kg
total kc = 1.081 $/kg

The total cost, including the proportional tool cost can be expressed as

KT ¼ kc mc þ kT

where

KT
kT ¼
50nc Lcm

KT is the tool cost, 50ncLcm is the total length of welded bars for 50 trucks (one
year production).
The results of calculations are shown in Table 2.
The total cost of the Al deck plate optimum (10 cross members of I-profile)
Kmin = 209.09 + 341.18 = 550.27 $
and that of the optimum (10 cross members of welded I-profile)
Kmin = 146.42 + 100.98 = 247.40 $
The cost savings for one steel truck is 302.87 $ (55%).

www.TechnicalBooksPDF.com
24 K. Jármai and J. Farkas

8 Conclusions

In the case of a truck floor welded from Al alloy extruded profiles and a deck plate,
the systematic optimum design process can result in significant savings in mass and
cost compared to the original design. A cross-section is optimized consisting of an
extruded cross member and an effective part of the deck plate. The objective
function is the cross-sectional area, the design constraints relate to a fatigue stress
range of welded joints and local buckling of extruded profiles. Fabrication aspects
regarding the size limitations are also considered.
In addition to the loading by pallets in horizontal floor position the case of
distorted floor position is also taken into account, when a truck wheel is staying on
a curb. The bending moments arising in this position have been calculated on the
basis of experimental measurements of deflections.
Optimization shows that the thickness of deck plate can be decreased from 4.5 to
2.0 mm, the original number of cross members can be decreased from 14 to 10, and
the original cross member shape (RHS) can be replaced by I- or a C - profile having
optimum dimensions. These changes can result in 141 kg mass and 377.17 $ cost
savings for a truck structure.
It should be emphasized that, in spite of the torsion of the whole floor in the
second loading case, the cross members are loaded by bending, since the torsion is
restrained by longitudinal members and by the deck plate. In the case of torsion the
RHS profile would be, of course, more advantageous than the open profiles.
Higher tool cost of the RHS for nc = 12 and 10 is caused by the large width of
the profiles, since the height is limited to 100 mm. It can be seen that the higher tool
cost does not significantly affect the result.
Using a welded steel deck plate and transversal stiffeners, one can make opti-
mization on the same way. In spite of the mass increment comparing to the alu-
minium optimum, using steel elements one can reduce the total cost of the structure
significantly, with 55%, although for a vehicle the mass is significant in fuel
consumption.

Acknowledgements The research was supported by the Hungarian Scientific Research

Fund OTKA T 109860 projects and was partially carried out in the framework of the Center of
Excellence of Innovative Design and Technologies in Vehicle, Mechanical and Energy
Engineering at the University of Miskolc within the EFOP-3.6.1-16-00011 “Younger and
Renewing University – Innovative Knowledge City – institutional development of the University
of Miskolc aiming at intelligent specialisation” project implemented in the framework of the
Szechenyi 2020 program. The realization of this project is supported by the European Union,
co-financed by the European Social Fund. The second author Prof. József Farkas† has recently
passed away. With this paper we remember for his 6 decades long valuable scientific work on steel
structure optimization.

www.TechnicalBooksPDF.com
Truck Floor Design for Minimum Mass … 25

References

1. DIN 1725–1983. Aluminiumlegierungen. Knetlegierungen. https://www.beuth.de/en/ standard/

din-1725-1/2650507
2. DIN 1748–1981. Strangpressprofile aus Aluminium und Aluminium-Knetlegierungen—
Zulässige Abweichungen. www.beuth.de/en/standard/din-1748-4/928185
3. FarkasJ, Jármai K (2013) Optimum design of steel structures, Springer Verlag, Heidelberg
288p. ISBN 978-3-642-36867-7. http://dx.doi.org/10.1007/978-3-642-36868-4
4. Farkas J, Jármai K, Dúl R (2001) Minimum cost design of a truck floor welded from
aluminium-alloy profiles, Welding in the World, Pergamon Press, Vol 45, No 9–10, pp 19–22.
ISSN 0043-2288
5. Hobbacher A. IIW Recommendations for fatigue design of welded joints and components.
IIW-doc. IIW-1823–07, ex XIII-2151r3-07/XV-1254r3-07
6. Eurocode 3. Part 1.1 (2005) Design of steel structures. General rules and rules for buildings.
European Committee for Standardization. Brussels
7. Rosenbrock HH (1960) An automatic method for finding the greatest or least value of a
function. Comput J 3(3):175–184. doi:10.1093/comjnl/3.3.175
8. London Metal Exchange http://www.lme.com/metals/ Accessed 17 Sept 2016

www.TechnicalBooksPDF.com
Theoretical and Parametric Investigation
of an Automobile Radiator

Máté Petrik, Gábor Szepesi, Károly Jármai and Betti Bolló

Abstract Automotive radiator is one of the most important devices of the engine
cooling system. The function of this equipment is to remove heat from the engine
and to keep the engine operating at the most efficient temperature. Nowadays, in the
automotive industry, one of the most important project is decreasing the mass. This
chapter focuses on calculation and optimization of finned-tube heat exchanger using
several methods.

1 Introduction

Automotive radiators are heat exchangers, which used for cooling internal com-
bustion engines. These engines are cooled by circulating engine coolant liquid
through the engine block where it is heated, and through this radiator, where it is
losing heat to the surrounding air. This coolant circulates from the tubes to the
engine block by the coolant-pump. The air forced across the radiator’s core, which
forced by a fan or by the motion of the vehicle. This air warms up by the coolant,
which temperature will be decreased. With the use of the fins, an extended surface
is obtainable. This radiator is a surface heat exchanger, so with this extended area
the performance will be higher [1].
The cooling system could be divided into two parts: the heat exchanger and the
air flow management components (such as pipes, water pump, thermostat, fan).
There is cooling performance difference between a car with a 77 kW power and a

M. Petrik (&)
G. Szepesi
K. Jármai
B. Bolló

University of Miskolc, Miskolc, Hungary
e-mail: vegypet@uni-miskolc.hu
G. Szepesi
e-mail: szepesi@uni-miskolc.hu
K. Jármai
e-mail: jarmai@uni-miskolc.hu
B. Bolló
e-mail: aramzb@uni-miskolc.hu

© Springer International Publishing AG 2017 27

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_3

www.TechnicalBooksPDF.com
28 M. Petrik et al.

truck with 340 kW. Due to limited space at the front of the engine, the size of the
heat exchangers is restricted and cannot be increased. So in case of higher cooing
performance, change of the radiator is not enough, investigated of the other parts of
the system (fan, fan shroud, coolant pump) is necessary. So, the optimization of an
automobile radiator includes the investigation of the single parts and the analysis of
interaction between them [2].
According to E. Sany, the performance of the radiator is the function of the inlet
temperature and the inlet velocity. Different tube rows and various coolant mass
velocities have been analysed in [3]. These results show that the increasing number
of tube rows, the performance of the radiator will be increased, but the pressure
drop of the device will increase too. The first result is useful, but the free space in
front of the device is limited. The second is harmful, because much stronger than it
must be used and deposits can be formed.
However, a change of the coolant could increase the performance too. In this
chapter, the coolant is water, but in the industry, there are a lot of other coolants are
used. Different thermal fluids have been compared in [4]: water, ethylene glycol and
propylene glycol aqueous solutions at 30, 40 and 50%. Their results show, that
from these the best coolant is the water. In the function of the coolant flow the
difference between the first and the second best performance could be approxi-
mately 2 kW too. This difference is not negligible. In this chapter, the coolant is
water for the best performance.

2 Calculation of the Automotive Radiator—Theorems

As all the other heat exchangers, the performance of the radiator is calculated by the
laws of the thermodynamics and criterial equations. The energy from the temperature
difference in case of the air and the coolant must be equal to the radiator performance.
These devices are cross-flow structures, where air is the outer and coolant is the
inner medium. The tubes were finned to improve the heat transfer. With an
increased heat transfer area, the performance will be bigger.
There are several theorems to calculate these finned tubes and shows differences
between the theories, which are described below.

2.1 Theorem of Fábry

According to György Fábry, the heat transfer coefficient of finned tubes depends on
the outer heat transfer coefficient, the heat transfer areas and the parameters of the
fins [5]. The fins disturb the heat convection between the air and the surface. The fin
efficiency is the quotient of the disturbed and original heat transfer coefficient.

www.TechnicalBooksPDF.com
Theoretical and Parametric Investigation … 29


AR A  AR
h ¼ h0 f b b þ ; ð1Þ
A A

where
h0 heat transfer coefficient of the unfinned tubes [W/m2K],
AR area of the fins [m2],
A the whole finned area [m2]
The f parameter can be calculated with these formulae:

0:63
l
f ¼ 1  0:18 ; ð2Þ
tR

where
l height of the fin [m],
tR gap between the fins [m]
The bb is the efficiency of the fins, which depends on the v parameter.
Calculation of v is:
sffiffiffiffiffiffiffiffiffiffi
2fh0
v¼h : ð3Þ
k R dR

The kR: heat conductivity of the fins material [W/mK] and dR: wall thickness of
the fins [m].
The overall heat transfer coefficient of the heat exchanger:


1 1 A d 1
¼ þ þ : ð4Þ
U h A1 k h1

More fins mean higher heat transfer area, but lower heat transfer coefficient. The
h1 is the heat transfer coefficient inner side of the tube [W/(m2K)] which calculated
in the function of the Re-number with these equations:
 0:8
0:19 Re
Pr dL
Nu ¼ 3:66 þ  0:467 ; when Re\2300 ð5Þ
1 þ 0:117 Re
Pr Ld
(
2=3 )
d
Nu ¼ 0:023 1 þ Re0:8
Pr0:4 ; when Re  2300 ð6Þ
L

www.TechnicalBooksPDF.com
30 M. Petrik et al.

2.2 Theorem of Perry

The heat transfer coefficient between the fins in the function of the face velocity of
the air, the outside diameter of the tubes and the center-to-center spacing is cal-
culated by directly according to this method. The heat transfer coefficient [6]:

0:6
v0:6
f p0
hf ¼ 5:29 0:4 0 : ð7Þ
D0 p  D0

The vf: the face velocity of the air [m/s], D0: outside parameters of the tubes [m],
p′: center-to-center spacing [m]
To the fin efficiency, two parameters must be calculated:
sffiffiffiffiffiffiffiffiffi
hf p f
m¼ ; ð8Þ
af k

where pf: perimeter of the fin [m] and af: cross-section area of the fin [m2].
The X must be calculated to determine the heat transfer coefficient:

tanhðmbf Þ
X¼ ; ð9Þ
mbf

where bf: height of the fin [m]. The heat transfer coefficient is similar to the
previous:


1 1 A d 1
¼ þ þ : ð10Þ
U Xhf A1 k h1

The appropriate Nusselt-equations to calculate the inner side heat transfer are:

0:0668
Ld Re
Pr
Nu ¼ 3:66 þ  2=3 ; when Re\2300 ð11Þ
1 þ 0:04 Re
Pr
Ld

2 ðRe  1000Þ Pr
f
Nu ¼ f 0:5  2=3  ; when Re  2300 ð12Þ
1 þ 12:7 2 Pr 1

where f is the friction factor: f ¼ 0:25
ð0:790
ln Re  1:64Þ2 .

2.3 Cengel’s Theorem

This theorem shows a lot of similarities with the previous two theories: the heat
transfer of the air side calculated by criterial equations (such as Fábry), but the

www.TechnicalBooksPDF.com
Theoretical and Parametric Investigation … 31

parameter in the fin-efficiency equations similar to the Perry-theorem. The

fin-efficiency as a function of this parameter is shown in a graph [7].
The Nusselt-number equation in case of forced convection over circular cylin-
ders in cross flow is:

Nuk ¼ 0:248
Re0:612
Pr1=3 : ð13Þ

The hydraulic diameter for the calculation of the Nusselt- and Reynolds-num-
bers [8]:

4
W
Afree
Dh ¼ ; ð14Þ
Aht

where W: the width in the orientation of flow [m], Afree: the flow section of the air
[m2] and Aht: the whole heat transfer area of the radiator [m2]. This type of
hydraulic diameter must be used because it is a non-circular cross-section with inner
flow breaker parts (the finned tubes). The velocity of the air is equal to the velocity
of the automobile.
The parameter for the fin-efficiency:

rffiffiffiffi
1 h
f¼ lþ t : ð15Þ
2 kt

The l is the length of the fins [m], and t: wall thickness of the fins [m].
The efficiency depends on this f value. An approximation curve was used in the
calculations.
The heat transfer coefficient calculated by:


1 1 A d 1
¼ þ þ : ð16Þ
U fhf A1 k h1

The criterial equations for the inner heat transfer are:

 
0:065
DL Re
Pr
Nu ¼ 3:66 þ  2=3 ; when Re\2300 ð17Þ
1 þ 0:04 Re
Pr
DL

Nu ¼ 0:125
f
Re
Pr2=3 ; when Re  23000; ð18Þ

f ¼ ð0:790
ln Re  1:64Þ2 : ð19Þ

www.TechnicalBooksPDF.com
32 M. Petrik et al.

3 Comparisons of the Theorem’s Results

The previous methods were used to calculate a commercially available cross-flow

car cooler. Parameters are shown in Table 1.
The results in Table 2, in case of steel structural material, are as follows:
Table 2 shows, that the calculation methods bring the same results. As a real
radiator is calculated, the outlet coolant temperature is a technical data. Increased
the performance this temperature will be decreased, which is the main goal of the
following calculations. When the fins are made of aluminium, the results with
the three theorems are according to Table 3.
There is a four-time ratio between the heat conductivities of the steel and the
aluminium. However, in case of the Fábry- and Perry-theorem the performance
does not significantly changed. The Cengel-method is sensitive to the change of the
material, but the others are not. The fin efficiency has the most considerable dif-
ference: in the first case is changed between 0.2 and 0.99, but in the other case
changed in a much smaller interval. If the fin material is copper it causes the largest
heat conductivity, which is about 386 W/(mK) according to Table 4.

Table 1 Basic data and used symbols of a commercially available cooler

Length of the radiator L 378 mm
Height of the radiator (length of tubes) H 378 mm
Width of the radiator W 34 mm
Outer diameter of the tubes Dout 8 mm
Wall thickness of tube ttube 1 mm
Numbers of tubes ntube 20 pieces  2 lines
Heat conduction of the tubes ktube 50 W/mK
Wall thickness of fins tfin 0.1 mm
Length of the fins lfin 18.9 mm
Spacing between the fins sfin 1.3 mm
Numbers of fins nfin 270 pieces/tube
Heat conductivity of the fins kfin 50 W/mK
Inlet air temperature Tairin 20 °C
Inlet coolant temperature Tcoolin 90 °C
Face velocity of air vair 70 km/h

Table 2 Results in case of steel structural material

Theorem Performance (W) Outlet coolant temperature Outlet air temperature
(°C) (°C)
Cengel 11,538 61.58 26.8
Fábry 11,155 62.52 26.57
Perry 11,294 62.18 26.66

www.TechnicalBooksPDF.com
Theoretical and Parametric Investigation … 33

Table 3 Results in case of aluminium structural material

Theorem Performance (W) Outlet coolant temperature (°C) Outlet air temperature (°C)
Cengel 12,451 59.33 27.78
Fábry 11,264 62.26 27.04
Perry 11,537 61.59 27.21

Table 4 Results in case of cooper structural material

Theorem Performance (W) Outlet coolant temperature (°C) Outlet air temperature (°C)
Cengel 12,652 58.84 27.9
Fábry 11,279 62.22 27.05
Perry 11,548 61.56 27.21

Table 5 Results in case of three structural materials

Fin material Performance (W) Mass (kg) Performance/mass (W/kg)
Steel 11,538 5.83 1940.27
Aluminium 12,451 2.00 6212.58
Copper 12,652 6.65 1902.17

Based on these tables can be obviously determined that better material heat
conductivity and fin efficiency cause higher performance. The sensitivity analysis is
calculated by the Cengel-method and the results shown below.

4 Sensitivity Analysis, Optimization

The heat exchanger performance depends on the structural material. However, these
materials have different density and mass. When the geometry is given (such in the
previous calculations), this performance and mass ratio are comparable as it is
visible in Table 5.
The steel is the less expensive, but has the lowest heat conductivity, so it causes
a large equipment. At a constant performance, the smallest device made of copper,
but it has a much higher price. The values of the aluminium are between the values
of steel and copper, but this ratio is three-times higher. So consider the price of the
radiator, the aluminium is the optimal choice.
In [3], they got similar results: the copper is the best choice, but the aluminium
has an approximately 4–5% less performance than the copper.
From this point, this performance/mass ratio change will be investigated in the
function of the change of the parameters. These parameters are the number of tubes,
the size of the tubes, number of fins and the width of the cooler.

www.TechnicalBooksPDF.com
34 M. Petrik et al.

4.1 Preliminary Calculations

In this calculation, the local optimum value is investigated in the function of

fin-parameter, tube size and numbers of tube-rows. The tube sizes are standard
values (Ø6  1 mm, Ø8  1 mm and Ø10  1 mm), and the rows are 1, 2 or 3.
The spacing between the fins are changed in the function of the fin-parameter.
As it can be seen in Fig. 1, when the number of rows increasing, the perfor-
mance and performance/ratio values are increasing too. There are local maximum
values of the fin-parameter of 250–300 values. So the increasing the fin parameter
the performance will be increased, but with it, the mass will more increasing, so
their ratio will be decreasing. However, in case of 1 row the radiator cannot satisfy
the required performance.
In Fig. 2, in case of Ø8  1 mm tubes, the performance is increasing, but the
mass is also increasing too, so the investigated performance/mass values are
smaller, then the first case. The Ø10  1 mm tubes are also investigated, but this
decreasing is observed. The decrease in the performance is because of the
increasing of the coolant flow area. The mass velocity of the coolant is constant, but
the free flow area is growing (the diameter and the number of tubes rises). So the
velocity of the water is reduced. This lower velocity causes lower Reynolds
number, which will affect lower heat transfer coefficient. The local maximum point
is shown Fig. 3 in the function of the fin-parameter.
From the calculations, this maximum point is at the 272 1/m fin-parameter. In
this case there are 102 fins on the tube, and the spacing between them is 3.64 mm at
this width. Any of the parameters are changed, this local maximum point will vary.

19000
Performance/mass [W/kg]

17000
Performance [W]

15000
13000
11000
9000
7000
5000
200 250 300 350 400 450 500
Fin parameter [1/m]
Performance (1 row) Q/m (1 row) Performance (2 row)
Q/m (2 row) Performance (3 row) Q/m (3 row)

Fig. 1 Performance of radiator from Ø6  1 mm tubes

www.TechnicalBooksPDF.com
Theoretical and Parametric Investigation … 35

19000

Performance/mass [W/kg]
17000

Performance [W]
15000
13000
11000
9000
7000
5000
200 250 300 350 400 450 500
Fin parameter [1/m]
Performance (1 row) Q/m (1 row) Performance (2 row)
Q/m (2 row) Performance (3 row) Q/m (3 row)

Fig. 2 Performance of radiator from Ø8  1 mm tubes

13600
13598
Performance/mass [W/kg]

13596
13594
13592
13590
13588
13586
13584
250 255 260 265 270 275 280 285 290
Fin parameter [1/m]

Fig. 3 Local maximum point in case of 3 rows and Ø6  1 mm tubes

4.2 Effect of the Radiator Width

The width is changed between 34 and 56 mm. Figure 4 shows the results:
Figure 4 shows that a wider radiator is better, but the optimum point move to the
smaller fin-parameters. So the free space in the car defines the maximum built-in
width. This free space is a basic planning data. This chapter shows a modified
geometry where larger performance can be reached with lower air velocity in
Table 6.
Whit this geometry, according to Cengel-method calculation the requested
performance (approx. 12 kW) can be reached with 16 m/s air velocity, but the
original heat exchanger needs more air velocity to produce this performance.

www.TechnicalBooksPDF.com
36 M. Petrik et al.

13800

Performance/mass [W/kg]
13600
13400
Performance [W] 13200
13000
12800
12600
12400
12200
100 150 200 250 300
Fin parameter [1/m]
38 44 50 56

Fig. 4 Effect of the width to the radiator performance

Table 6 Original and the Original Optimized

modified heat exchanger
geometry Width 34 mm 56 mm
Tube size Ø8  1 Ø6  1
Number of tubes 2  20 pieces 3  25 pieces
Spacing between fins 1.3 mm 6.77 mm
Length of fins 18.9 mm 15.12 mm
Number of fins 270 pieces 56 pieces
Mass 1.68 kg 1.10 kg

5 Summary

This chapter point out this area is still an interesting topic in the field of heat
exchangers. Several calculation methods were introduced to calculate the perfor-
mance of the car radiators which is a typical cross-flow air-liquid finned heat
exchanger. As this chapter showed, a lot of parameters can affect the performance,
but the main parameters are the radiator width and the fin parameter. Two calcu-
lations presented where the performance change can be calculated in the function of
these parameters. A new construction is also demonstrated where lower air speed is
enough to reach the acceptable performance. It would be very important in case of
electrical cars where the noise is critical.

Acknowledgements The research was supported by the Hungarian Scientific Research

Fund OTKA T 109860 project and was partially carried out in the framework of the Center of
Excellence of Innovative Vehicle, Mechanical and Energy Engineering Design and Technologies
at the University of Miskolc.

www.TechnicalBooksPDF.com
Theoretical and Parametric Investigation … 37

References

1. Carl M, Guy D, Leyendecker B, Miller A, Fan X (2012) The theoretical and experimental
investigation of the heat transfer process of an automobile radiator. In: 2012 ASEE gulf
southwest annual conference, April 4–6, El Paso, Texas, vol 1 (128), pp 1–12
2. Esmaeili Sany AR, Saidi MH, Neyestani J (2010) Experimental prediction of Nusselt number
and coolant heat transfer coefficient in compact heat exchanger performed with e-NTU method.
J Engine Res 18:62–70
3. Charyulu DG, Singh G, Sharma JK (1999) Performance evaluation of a radiator in a diesel
engine—a case study. Appl Therm Eng 19(6):625–639
4. Oliet C, Oliva A, Castro J, Pérez-Segarra CD (2007) Parametric studies on automotive
radiators. Appl Therm Eng 27(11):2033–2043
5. Fonyó Z, Fábry G (1998) Vegyipari művelettani alapismeretek. Nemzeti Tankönyvkiadó
6. Green DW, Perry RH (2008) Perry’s chemical engineers’ handbook
7. Cengel YA (2003) Heat transfer—a practical approach, 2nd edn.
8. Amrutkar PS, Patil SR, Shilwant SC (2013) Automotive radiator—design and experimental
validation 3(4):1–10

www.TechnicalBooksPDF.com
Past and Present: Teaching and Research
in Vehicle Engines at the University
of Miskolc

Szilárd Szabó, Péter Bencs and Sándor Tollár

Abstract The Department of Fluid and Heat Engineering, established as the

Department of Machine Operation in 1952, hosts research and teaching in engines
for vehicles. In the first two decades after the foundation of the department, besides
engines in vehicles, steam engines still played an important role as they were used in
locomotives. At that time the department owned several dynamometers and besides
teaching about engines, the department carried out research and development for the
national production of engines. After a decade of transition, in the second half of the
1990s the production of internal combustion engines started to flourish again. This
was manifested first in teaching, but from 2010 in research as well. The replacement
of the old laboratory equipment with modern measuring instruments and engines and
the construction of a full departmental laboratory for engine diagnostics contributed
to this. All of these factors explain why the Department of Fluid and Heat
Engineering is in charge of teaching internal combustion engines in the new (starting
from 2016) Vehicle Engineering B.Sc programme.

1 Introduction

The Department of Machine Operation was established in the Faculty of

Mechanical Engineering at the Technical University of Heavy Industry by Decree
25 of 1951. The first head of department was Professor Alajos Lancsarics
(1903–1963) (Fig. 1).

S. Szabó (&)
P. Bencs
S. Tollár

Institute of Energy Engineering and Chemical Machinery, Department of Fluid
and Heat Engineering, University of Miskolc, Miskolc, Hungary
e-mail: aram2xsz@uni-miskolc.hu
P. Bencs
e-mail: arambp@uni-miskolc.hu
S. Tollár
e-mail: aramts@uni-miskolc.hu

© Springer International Publishing AG 2017 39

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_4

www.TechnicalBooksPDF.com
40 S. Szabó et al.

Fig. 1 Alajos Lancsarics,

founding professor of the
department

Since its foundation, the department has hosted research and teaching in engines
for vehicles. After its educational and research profile became established, the name
of the department was changed to the Department of Fluid and Heat Engineering by
Decree 52341/1965, MM on 15 March 1965.
The organisational structure of the Faculty of Mechanical Engineering and
Informatics was modified by the formation of institutes in the autumn of 2013,
leading to the establishment of the Institute of Energy Engineering and Chemical
Machinery. Two previously existing departments joined to form the institute, and
our activities go on within the Institute.

2 In the Beginning

Alajos Lancsarics was legendary for his enthusiasm for teaching machinery, and
especially heat engines. As the vice rector for financial affairs, it was his task to set
up and expand the equipment available for workshops and laboratories, which were
quite rudimentary in the beginning. He managed to provide the necessary back-
ground needed for teaching about internal combustion engines, despite the numerous
difficulties faced in the 1950s.
Heat engines were a major part of the content taught at that time, with a focus on
internal combustion engines and also steam engines (see Fig. 2), as their role in
locomotives was still important at that time.
It was mainly the tireless work and wide professional knowledge of Alajos
Lacsarics that contributed to the development of teaching materials for the newly
formed department (Fig. 3). He was enthralled by motorization, and in his opinion a
degree in mechanical engineering could not be granted to someone lacking a
thorough knowledge of these machines. In the laboratory classes for engines, he
arranged trips for pairs of students between Miskolc and Hatvan in a steam loco-
motive, or organised a special train to take all of the mechanical engineering
students on a visit to a power plant, using the Campus-Tiszapalkonya line. As early
as the 1950s, students were learning how to drive motorcycles (Fig. 4), automo-
biles, tractors and combines.

www.TechnicalBooksPDF.com
Past and Present: Teaching and Research in Vehicle … 41

Fig. 2 Prof. Lancsarics lecturing on heat engines in the 1950s

Fig. 3 Collection of problems for heat engines (left); hand-drawn lecture notes of a control unit
and a regulator, by A. Lancsarics (right)

www.TechnicalBooksPDF.com
42 S. Szabó et al.

Fig. 4 Students learning how to drive a motorcycle on campus

Fig. 5 Motor testbeds in the 1950s

In the first two decades, the department laboratories were located in the make-
shift premises earlier inhabited by the prisoners who built the first buildings of the
university. In addition to housing various types of vehicles, there was also an
engine test lab with a testbed. That the members of the department had their
photograph taken with this equipment (Fig. 5) testifies to their strategic importance.
During those first two decades, the department was involved not only in teaching
but also in R&D on behalf of Hungarian motor manufacturers. These activities also
supported the research progress made by lecturers. An example is the 1966 thesis
for the university doctor’s degree written by György Vida on the factors deter-
mining heat transfer in the cylinders of a diesel engine.
From the mid-1960s the teaching and research profile of the department altered
somewhat, under the leadership of Tibor Czibere, with hydraulic machines gaining
more emphasis among the turbomachines. Naturally internal combustion engines
remained a focus of interest, as shown by the fact that half of the space in the

www.TechnicalBooksPDF.com
Past and Present: Teaching and Research in Vehicle … 43

modern laboratory (established in 1969) was given over to the engine testbeds,
while the other half was used for the teaching of and research on hydraulic
machines (Figs. 6 and 7).
In the 1990s, most of the motor manufacturing in Hungary took place on a
licensing contract basis, and the amount of R&D was quite limited. During this
period the main focus for engines was in teaching, while hydraulic turbines were a
topic not only in the classroom but also for research.

Fig. 6 The new department laboratory (1969), with testbeds on the left and underground water
reservoirs and hydraulic machines on the right

Fig. 7 A modern testbed with dynamometer (1969)

www.TechnicalBooksPDF.com
44 S. Szabó et al.

With the settlement of multinational companies in Hungary in the new millen-

nium, the production of internal combustion engines started to flourish in Hungary.
At the same time gas engines and gas turbines came to the front in energy pro-
duction. That is why the demand for engineers specialising in these fields has
increased again. Awareness of these trends motivated the department to revise the
curriculum and update the equipment needed for teaching and research. Within the
subjects of Turbomachines, Power Stations and Systems of Power Engineering
within the B.Sc programme in Energy Engineering special attention was given to
heat engines, steam turbines, gas turbines and gas engines. In the area of internal
combustion engines educational equipment was first procured: a 90 hp 1.4 Opel
Astra G petrol engine of 1997, a 2.0 HDI (High Pressure Direct Injection) Peugeot
diesel engine of 2002, vibration diagnostic equipment called EASY-VIBER, a
Testo type exhaust gas analyser and Würth type automotive diagnostic equipment.
Using this equipment, new courses were launched such as Internal Combustion
Engines and Engine Diagnostics.
As the next step, in 2010 an engine test stand for the engine braking of single
cylinder petrol and diesel engines produced by GUNT was purchased (see Fig. 8).
The test stand was produced as equipment for engineering education but can also be
used for doing some simple research. For example, using this equipment it became
possible to investigate fuel consumption [1] or the use of additives for reducing fuel
consumption. Using the apparatus, students investigated the effect of auxiliary
devices within complex assignment and final thesis projects.

Fig. 8 An engine test lab system with a drive motor and dynamometer produced by GUNT
(2010)

www.TechnicalBooksPDF.com
Past and Present: Teaching and Research in Vehicle … 45

3 Nowadays

The largest development took place in 2012. The research and teaching laboratory
was established in two rooms of the departmental laboratory for engine diagnostics.
The establishment of the laboratory was mainly supported by a grant for infrastruc-
tural development. The engineering and technological design and the implementation
of the project were carried out by Energotest Ltd. [2]. The name of the laboratory
became the Lancsarics Engine Test Lab after the founder of the department.
The engine test room was established using existing machines. The 2.0 TD
Common-rail diesel engine used for the first experiments was awarded to the
University of Miskolc by Audi Hungaria Motor Ltd. The tested engine (Fig. 9) was
placed in a soundproofed room separated from a control and teaching cabin.
Soundproof fire doors between the two rooms ensure visibility and accessibility.
The fixing of the engine to the base of the machine and positioning it to the
dynamometer is ensured by specially developed engine holder palettes. The

Fig. 9 Interior of the engine test laboratory

www.TechnicalBooksPDF.com
46 S. Szabó et al.

connection between the engine and dynamometer for the transfer of torque and
speed (of revolution) is ensured by a Cardan shaft with a very flexible rubber
element for absorbing torsional vibrations and equipped with fitting disks. The
8000 rpm dynamometer (maximum power 250 kW, maximum torque 1200 Nm) is
water-cooled eddy current dynamometer with impulse modulation control elec-
tronics. A 44 kW induction motor with frequency converter speed control is
attached to the engine brake side of the powertrain system, constituting a compound
braking unit developed by Energotest. The controllable induction motor enhances
the dynamics of the system substantially and also the effectiveness of the dynamic
tests. Also, the induction motor is capable of driving the test engine, hence the
engine operation can be avoided and tests can be carried out using an electric motor
to rotate the engine, ensuring the investigation of engine friction.
The environmental and technological boundary conditions are provided by a
preparatory and serving measuring system developed and built by Energotest with
the following elements:
• A preparatory system for liquid coolant that provides cooling of the test motor
using a water–water heat exchanger, built-in pipe system, a pump whose
parameters match those of the system, and control elements.
• A preparatory system for diesel fuel provides the fuel needed to drive the test
motor. The cooling of the fuel is carried out with a fuel–water heat exchanger.
• A preparatory system for the air intake of the motor provides the proper filtering
of the air needed for motor operation, with adjustable de-pressurization and
temperature control in the temperature range of 15–40 °C.
• An exhaust system removes the emitted fumes with corrosion-free piping and a
compensator. A test section allows emission measurements.
• A ventilation system for the room provides continuous ventilation of the lab,
with ventilators removing and introducing air, air ducts, fume hood, filtering
system and rain proofing elements.
• An accelerator unit ensures the control of modern E-gas systems as well as
conventional bowden cable systems.
• A CAN bus data acquisition system can be configured flexibly and the poten-
tially extendable version allows the measurement of motor and environmental
parameters up to 20 signals. It can also control the preparatory system and that
of the test engine. The whole system is computer controlled with Hungarian
language Energopower engine test bed software installed in an industrial
computer built into the rack-type control desk.
The up-to-date engine test laboratory is suitable for investigating energy pro-
cesses taking place in internal combustion engines and for carrying out diagnostic
and emissions tests on diesel engines for passenger cars and light commercial
vehicles, in use today or for future development, as well as for educational activities
related to the practical operation of motors.
With a dynamometer, during motor operation, it is possible to determine charge
characteristics, investigate part loads and test cycles defined by the user, measure

www.TechnicalBooksPDF.com
Past and Present: Teaching and Research in Vehicle … 47

fuel consumption, determine specific fuel consumption curves, and investigate the
effect of changes in environmental parameters on performance. In addition, with the
environmental system the effects of various types of motor oil and fuel can be
investigated, and harmful emissions can be analysed using an AVL emission meter
of “0” accuracy class.
Being able to rotate the engine by an electrical motor enables us to investigate
the friction at different speeds, and to carry out tests as used by manufacturers for
engine development. This is a fundamental measuring technique for the downsizing
experiments expected these days.
Since being put into operation the engine test bed has been engaged in three
large research projects when long tests for different purposes have been carried out
on diesel engines:
• 100-h test to check the laboratory operation at different engine operating conditions
• 800-h test of deposition and soot deposition
• 500-h test of motor wear and oil consumption.
All of these factors explain why the Department of Fluid and Heat Engineering
is in charge of teaching about internal combustion engines in the Vehicle
Engineering B.Sc programme being launched in 2016. This includes subjects such
as Internal Combustion Engines and Motor Vehicle Engine Diagnostics.
Research within the department on two doctoral-level research topics is currently
underway: one deals with experimental and numerical investigation of the exhaust
system and the other with the development of an alternative valve.
In addition to measuring devices and apparatus, the software package ANSYS
Fluent and AVL motor diagnostics software are available for numerical simulation.
Further development of the engine test laboratory is planned in order to
accommodate the latest models of engines, and we would like to expand the
research topics by purchasing equipment for vibration diagnostics in order to
investigate mechanical effects of vibration.

Acknowledgements The research was partially carried out in the framework of the Center of
Excellence of Innovative Design and Technologies in Vehicle, Mechanical and Energy
Engineering at the University of Miskolc.
The described article was carried out as part of the EFOP-3.6.1-16-00011 “Younger and
Renewing University – Innovative Knowledge City – institutional development of the University
of Miskolc aiming at intelligent specialisation” project implemented in the framework of the
Szechenyi 2020 program. The realization of this project is supported by the European Union,
co-financed by the European Social Fund.

References

1. Tollár S, Mátrai Z (2012) Investigation of the effect of different diesel fuels on operating
parameters of an engine. GÉP 62(9):57–60
2. Szilágyi G (2012) New dual-function dynamic engine test lab at the University of Miskolc.
GÉP 62(9):5–6

www.TechnicalBooksPDF.com
Alternating Current Hydraulic Drive
the Possibility of Applying
in the Automotive Industry

Tamás Fekete

Abstract For the drive technology tasks are used different system drives. One of
these drives is the well-known direct-current hydraulic drive (DCH). The hydraulic
drives can be classified among a new version of the new hydraulic drives: the
alternating-current hydraulic drive (ACH). The spread of the drive innovation
problems hinder, in turn it has many favourable features, which influence positively
the transmission properties of the drive. With the development of this type of drive
is dealt at the Department of Machine Tools the leadership of Dr. János Lukács
about 40 years. The subject in the field of several doctoral dissertations, patents,
scientific articles and student papers have been written. I got into the research work
within the framework of doctoral training. My task was within the topic of
alternating-current hydraulic drives: the transmission properties of the synchronous
alternating-current hydraulic drive. I would like to introduce on the Department of
Machine Tools that I built help of alternating-current hydraulic drive, this drive is
advantageous and disadvantageous features, knowing that they these transmission
properties can be used advantageously machines of the automotive industry.

Keywords Synchronous alternating-current hydraulic drive
Hydromotor

Hydrogenerator

1 Introduction

By all of the vehicles, used in other areas of technology, and by all of the implement
(not knitted track travelling) the fundamental problem is the design of the con-
trollability and the planning of his drive is necessary to the operating of his ensuring
movement equipment. The implementing of the drives may be using mechanical,
electrical, pneumatic and hydraulic energy is. The drives are from among operated
by technology listed introduced hydraulic drives I would like to introduce a drive,

T. Fekete (&)
University of Miskolc, Miskolc, Hungary
e-mail: tamas.fekete@uni-miskolc.hu

© Springer International Publishing AG 2017 49

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_5

www.TechnicalBooksPDF.com
50 T. Fekete

which is based on the positive properties of drive technology in the automotive

industry, movement tasks of mobile appliances can be used.
According to the electrical analogy of the hydraulic drives can be divided into
two main groups: direct-current hydraulic drives (DCH) and alternating-current
hydraulic drives (ACH) [1].
By direct-current hydraulic drives, the energy transfer fluid (for example oil)
between pump and motor(s) does unconfined movement. In case of direct-current
hydraulic drives, the exotherm is substantial because of the hydraulic resistance.
This exotherm must be drained out of the system. The draining of generating heat
can carry out with the help of a heat exchanger or big size holder.
By the alternating-current hydraulic drive the oil in the phasepipes performs
small amplitude alternating movement, therefore smaller in the phase space the
thermogenesis.
In phase transmission are oil by alternating-current hydraulic drive makes an
alternative motion. Therefore, temperature dissipation is lesser, because the para-
bolic speed distribution in phase transmission can’t develop for reciprocating
motion. The alternating-current drive can be divided into two groups: synchronous
and non synchronous.
The ACH drives have two types. The alternating current synchronous drive
(S-ACH in Fig. 1), the alternating current nonsynchronous drive (ACHM) (and the
alternating current asynchronous drive (A-ACH)). The ACH drives have three main
units: the alternating current hydrogenerator (ACG), the alternating current
hydromotor (ACM) and the phase space (phasepipes) [2].

ACG ACM
PHASEPIPES

EXCENTER 1 EXCENTER 2

Fig. 1 The experimental drive made by me of the synchronous alternating-current hydraulic drive
(S-ACH)

www.TechnicalBooksPDF.com
Alternating Current Hydraulic Drive the Possibility … 51

ACG PHASEPIPES

ACM

EXCENTER 1 EXCENTER 2

Fig. 2 Schematic outline of the synchronous alternating-current hydraulic drive (S-ACH)

The ACG of the pilot model has an eccentric exciting element. The shaft of the
ACG is driven by a direct current hydro motor. By applying a direct current hydro
motor simply can be enable the frequency controlling by the setting of the driving
flow. We can also set the amplitude of the phase-flow with a double eccentric wheel
when the drive is standing [3].
The experimental drive made based on the schematic outline (Fig. 2).

2 The Synchronous Alternating-Current Hydraulic Drive

(S-ACH)

Contrast to direct-current drive, one advantage of alternating-current hydraulic

drive is that, big size holder doesn’t need. A central hydrogenerator (Fig. 3) gen-
erates the pulsating liquid flow (ACG), and propulsion of hydromotors
(ACM) (Fig. 4) on given equipment with pulsating liquid flow can be feasible
separately or in combination.
The hydrogenerator driven takes place by an external drive, which can be by an
electric motor, gasoline or diesel internal combustion engine or a hydromotor. In
my case the drive takes place by direct-current hydromotor. The direct-current
hydromotor drive is shown in the Fig. 3.
The generator is an energy source in the chain of the energy transfer. The main
parts of the generator are: the phase cylinder, the implied phase pistons and the
motion mechanism of the phase pistons. These all cause the generation of the
hydraulic sections, which determine the liquid flow. In case of synchronous drive
case, the design of both the ACG and ACM is the same. In the Fig. 4 we can see the
hydromotor with the load module.

www.TechnicalBooksPDF.com
52 T. Fekete

DIRECT-CURRENT ACG
HYDROMOTOR

CLUTCH

Fig. 3 The hydrogenerator of the synchronous alternating-current hydraulic drive (ACG) with the
direct-current hidromotor

TORQUEMETER ACM

LOAD MODULE

Fig. 4 The hydromotor of the synchronous alternating-current hydraulic drive (ACM) with the
load module

From knowledge of the theorem of movement of excenter we can determine by

the movement of the pistons. Examine the movement of excenter and that is the
impact of the movement of the pistons. Based on the Fig. 5 we can determine that
the value of q is differentially associated changing of the angle u.
The theorem of movement of the alternating-current hydraulic drive it can be
calculated as follows:

www.TechnicalBooksPDF.com
Alternating Current Hydraulic Drive the Possibility … 53

Fig. 5 The effect of the movement of the excenter of the piston displacement

r ¼ R þ e
cosu; ð1Þ

where r—a selected point of the excenter of movement distance from the pivot point
[mm], R—radius of the excenter [mm], e—the eccentricity [mm], u—[°].
In case of 120° offsets, why the pistons in case of three-phase spaced relative to
one another 120°:
9
r1 ðuÞ ¼ R þ e cos u >

=
r2 ðuÞ ¼ R þ e cos u þ 2p : ð2Þ

3
4p ;
>
r3 ðuÞ ¼ R þ e cos u þ 3

So the displacement of the pistons is:

9
s1 ¼ e½1  cosðx tÞ >
=

 
s2 ¼ e 1  cos x t þ 2p
ð3Þ

3 
4p ;
>
s3 ¼ e 1  cos x t þ 3

where s—displacement of piston [mm].

The alternating-current hydraulic drive reach at creep loads. These loads are
ohmic inductive and capacitive hydraulic resistances. I disregard for these sim-
plicity trial mode.

www.TechnicalBooksPDF.com
54 T. Fekete

The characteristic curve of the liquid flow, generated by ACG, is all most same
~ i ði ¼ 1; 2; 3Þ phase liquid flow can be calculated with the
like sine function. The Q
following equation:
 
2p
Qi ¼ Q0
sin x t þ k
; ð4Þ
3

The absolute value of the liquid flow isc

Q0 ¼ A
e
xg
; ð5Þ

where A—surface of the piston [m2], e—the eccentricity [mm], xg—angular velocity
of the generator [1/s].
The law of motions of the pistons in the three hydraulic cylinders is
9
xg1 ¼ r cos xt >
=


xg2 ¼ r cos xt  2p ; ð6Þ

3 >
4p ;
xg3 ¼ r cos xt  3

The surface of the foreheads of the pistons is the same.

The
/h1 ¼ Lh1 Q1 ð7Þ

where /h —the hydraulic flux, Lh —the hydraulic induction factor, Q—volume flow
rate or liquid flow, and

m dxgl
/h1 ¼ ; ð8Þ
A2 dt

from these relations

m
Lh ¼ : ð9Þ
A2

With substitution Q the liquid flows are calculatedly:

9
Q1 ¼ Ar sinðx tÞ >

=
Q2 ¼ Ar sin x t  2p : ð10Þ

3
4p ;
>
Q3 ¼ Ar sin x t  3

The summing of the three liquid flows the following amount:

Q1 þ Q2 þ Q3 ¼ 0 ð11Þ

is got [4, 5].

www.TechnicalBooksPDF.com
Alternating Current Hydraulic Drive the Possibility … 55

The functioning of the hydraulic generator starts the alternating current of liquid
flows in the phase pipe, and these liquid flows forced the excenter disc of the motor
to rotate.
The essential condition Pfor operating the system is that summarise of the three
liquid flows must be zero ( Q ¼ 0). If the condition does not come true, pressure
increase will appear in the phase pipe.
Each phase space of the capacitive pressure changes can be determined:
Z
1

Dp1 ¼  xg a sin xg t þ dt
CH
Z  
1 2p
Dp2 ¼  xg a sin xg t þ  dt ð12Þ
CH 3
Z  
1 4p
Dp3 ¼  xg a sin xg t þ  dt
CH 3

The phase printing may not be less than zero value (the phase space may must be
tensioned, such as with pump). The phase pressures and knowledge of the torque
arm of the actuation torque can be calculated:

Mh ¼ Apc1 k1 þ Apc2 k2 þ Apc3 k3 : ð13Þ

3 Measurement

We conducted the experimental measurements with flexible phasepipes, which we

can see in Fig. 6. Based on the measuring arrangement the advantage of the flexible
phasepipes is that the drive can be transmitted in any way, disadvantage that the
capacitive character of the system is increased, so that the driving power is reduced.
The target of the measurement, we got to know to thoroughly the transmission
properties of the synchronous alternating-current hydraulic drive (S-ACH), context
of theoretical torque and speed (rotation) characteristics.
We can be determined with the measurement configurations main characteristics
of the drive, the rotation speed of the hydromotor and load capacity, and their
context using the appropriate measuring instruments.
Under the given circumstances must be examined which drive should choose to
purpose (flexible or rigid phasepipe). In this paper, we try to give an answer to this
question and we would like to demonstrate its characteristics and measurement
form and method, wherewith parameters can be controlled. For this purpose, with
measurements we would like to demonstrate the synchronous drive, it’s a imple-
mented method and characterized by state parameters (Fig. 7).

www.TechnicalBooksPDF.com
56 T. Fekete

Fig. 6 The measuring arrangement

1 PHASE

2 PHASE
- 3 PHASE

Fig. 7 The characteristics of the liquid flows in the phasepipes

4 Conclusion

From the present paper, it can be deduced, that the alternating-current hydraulic
drive is suitable automotive industry drive technology tasks can be used. This
technique gives the development opportunity, that the generator speed adjustability
of modern electrical engineering can be realized. Later, it should have a more
thorough examination of the effects of heat. Later, it should have a more thorough
examination of the heat effect.

www.TechnicalBooksPDF.com
Alternating Current Hydraulic Drive the Possibility … 57

Third consideration is the possibility of the incorporation of other alternating-

current hydraulic drive, e.g.: a hydrotransformer. We would with this hydrotrans-
formator increase the distance between the hydromotor and the hydrogenerator, but at
the same time the temperature can be reduced. We could be achieved with the
hydrotransformer using that a small generator a bigger size hydraulic motor (and vice
versa) to operate.

References

1. Fekete T (2014) The alternating current synchronous hydraulic drive. Annals of faculty of
engineering Hunedoara, Int J Eng 12:235–238
2. Lukács J, Erdélyi J (2005) Operation and construction issues of the AC hydro generator phase
pistons. Pneumatika, hidraulika, hajtástechnika, automatizálás. Vol. IX. pp. 60–63 (in
Hungarian)
3. Erdélyi J, Fekete T, Lukács J (2008) The constructional and operational characteristics of
contraction cylinder. Pneumatika, hidraulika, hajtástechnika, automatizálás Vol. XII. pp. 3–5
(in Hungarian)
4. Pattantyús ÁG (1951) Practical Fluid mechanics. Budapest, Tankönyvkiadó (in Hungarian)
5. Ponomarjov Sz D (1966) Strength calculations in mechanical engineering. Vibrations SHOCKS.
Budapest, Műszaki Könyvkiadó (in Hungarian)

www.TechnicalBooksPDF.com
Comparative Destructive
and Non-Destructive Residual Stress
Measuring Methods for Steering Rack
Bar Semi-Product

József Majtényi, Viktor Kárpáti, Márton Benke

and Valéria Mertinger

Abstract It is well known that residual stress is introduced within solid materials
during many types of processing methods, including heat treatments, machining,
grinding, casting, etc. The type and magnitude of the formed stress state can be
various depending on the type and conditions of the treatment and the geometry of
the sample. The presence of residual stress can either be harmful or useful. If
undesired residual stress is arisen within a machine component during its manu-
facturing steps, it can lead to deformation. Since the geometry of automotive
components must be kept strictly within tolerances, more and more attention is
given to the importance of residual stress in the field of the automotive industry.
Many methods have been developed to characterise the residual stress. In the
present study, the results of a destructive and non-destructive residual stress mea-
suring methods for induction hardened automotive rack bar semi-products were
compared. The theory of non-destructive method associated with the distortion of
the crystal lattice, from which, residual stress can be quantitatively calculated, while
the destructive method that we have developed specially to test this product at the
Lech-Stahl Veredelung GmbH is very fast, suitable for qualification, and easily
inserted into the manufacturing line.

J. Majtényi (&)
V. Kárpáti
M. Benke
V. Mertinger

Institute of Physical Metallurgy, Metalforming and Nanotechnology,
University of Miskolc, Miskolc, Hungary
e-mail: jozsef.majtenyi@lech-stahlveredelung.de
V. Kárpáti
e-mail: cirkonium2000@gmail.com
M. Benke
e-mail: fembenke@uni-miskolc.hu
V. Mertinger
e-mail: femvali@uni-miskolc.hu

© Springer International Publishing AG 2017 59

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_6

www.TechnicalBooksPDF.com
60 J. Majtényi et al.

1 Introduction

Residual stress is arisen in solid state materials due to various types of manufac-
turing methods such as heat treatments, plastic deformation, machining, grinding,
casting, etc. The type (tensile/compressive) and magnitude of the stress state can be
extremely various depending on the characteristics of the processing method,
conditions, the geometry and material properties of the component. The occurrence
of residual stress can either be harmful or useful [1]. The formed residual stress
within a component can harmfully affect its geometry during many types of
manufacturing steps, therefore, more and more attention is payed to the phe-
nomenon of residual stress in the field of the automotive industry. Nowadays, many
methods are available to describe the residual stress state of a component. The
majority of these examination methods are based on the principle that mechanical
stress is always balanced within a component. However, if material is removed
from the component, the original stress state will undergo a change which results in
the loss of the previous geometry, that is, the component undergoes macroscopic
deformation. Despite these macroscopic deformation-based examinations are cap-
able for qualification or even classification, none of them provide accurate stress
values. Diffraction based techniques are based on the strain of the crystal lattice,
from which, residual stress can be calculated using mechanical constants of the
examined material [1, 2].
In the present manuscript, the residual stress states of automotive steering rack bar
semi-products produced by Lech-Stahl Veredelung GmbH were investigated. The
main processing steps of the steering rack bars are induction hardening, annealing,
burnishing and polishing in a continuous processing line. It is known that the
developed stress state resulting from induction hardening can be various depending on
the conditions of cooling and component geometry [3–5]. Furthermore, if subsequent
processing steps follow, the final stress state of the component can be much more
complex. This study focuses on the stress states formed after quenching, tempering
and, in addition, to achieve higher stress values, stress was intentionally introduced
into the steering rack bars through direct water cooling after stress relaxation heat
treatments. The effect of quenching temperature on forming stress states was also
examined. The stress states were characterised using non-destructive (sampling-free)
X-ray diffraction and macroscopic deformation. The different examinations in prin-
ciple were carried out on the same steering rack bars for better comparison. The
correlation was searched for between the results of the two methods.

2 Experimental

Steering rack bar semi-product rods with diameter (D) of 26 mm and length of
500 mm made of 37CrS4 type steel were processed in a continuous manufacturing
line at Lech-Stahl Veredelung GmbH, in Oberndorf am Lech, Germany. The first

www.TechnicalBooksPDF.com
Comparative Destructive and Non-Destructive … 61

main manufacturing step was heat treated which three applied quenching temper-
atures, being 860, 880 and 900 °C, while annealing temperature was constant,
namely 720 °C. Afterwards, the rods were subjected to machining and polishing,
which were followed by stress relaxation heat treatment and final polishing. For
some rods, water cooling was applied after stress relaxation to induce residual stress
within the semi-products. The residual stress state of the rods was characterised
with non-destructive (sampling-free) X-ray diffraction method at the Institute of
Physical Metallurgy, Metalforming and Nanotechnology of the University of
Miskolc, Hungary with a Stresstech X stress 3000 G3R centreless X-ray diffrac-
tometer. The residual stress was calculated from the shift of the {211} reflection of
ferrite using Cr Ka radiation, Young’s modulus (E) of 210000 MPa and Poisson’s
ratio of 0.29. The reflections were recorded in w geometry and ± 3 tilting positions
in the −45°/45° tilting range [6]. Stress was measured in 5 points on each of the
eight generatrix of the rods Fig. 1. For one rod, stress was measured in a point
series 20 mm from each other on the eight generatrix. The stress was measured with
less scatter (error) than 50 MPa in every measurement point, therefore scatter
ranges are not marked in the figures.
If the true cause of bending is searched for, the difference between the stress
values of the opposing generatrix must be considered. This is true for pure bending
of a rod, where a tensile and a compressed genetratrix exists on the opposing sides
of the rod. If an additional homogeneous stress is applied to the whole rod, for
example, compressive stress, the stress value of the tensile generatrix decreases
while the stress of the compressed generatrix increases. Therefore, to measure the
true tendency of bending, stress asymmetry (r*) must be calculated as the absolute
value of the difference of stress values along the two opposing generatrix of the
rods. An example for the formula of stress asymmetry for generatrix “A” and “E” is
given in Eq. 1 [7].

dAE ¼ jdA  dE j ð1Þ

During macroscopic deformation examinations, the eccentricity measured

20 mm from the end of the rods. Afterwards, a 180 mm long section with a depth of
0.3xD was machined from the rods 20 mm from the end of the rods on the gen-
eratrix of the highest measured eccentricity Fig. 2. After machining, eccentricity
was again measured and the difference (D) of eccentricity before and after
machining was calculated to characterise macroscopic deformation.

Fig. 1 Residual stress measurement mesh of the examined rods

www.TechnicalBooksPDF.com
62 J. Majtényi et al.

3 Results and Discussion

3.1 Residual Stress States

Figure 3 shows the measured residual stress values of the rods after quenching from
(a) 880 °C and (b) 900 °C, prior to annealing. Average stress values are inserted.
The stress values are low, varying between −100 and 100 MPa. The average stress
values are *40 MPa. No differences can be observed between the stress states after
quenching from 880 and 900 °C.
Figure 4 shows the measured residual stress values of the rods after annealing.
Three quenching temperatures were used: (a) 860 °C, (b) 880 °C and (b) 900 °C.

Fig. 2 Macroscopic deformation measurement performed by Lehl-Stahl Veredelung GmbH

Fig. 3 Residual stress

distributions of the examined
rack bar rods quenched from
a 880 °C b 900 °C

www.TechnicalBooksPDF.com
Comparative Destructive and Non-Destructive … 63

Fig. 4 Residual stress

distributions of the examined
rack bar rods after annealing.
Quenching temperatures are
(a) 860 °C (b) 880 °C
(c) 900 °C

Average stress values are inserted. The rods are stress-free, the measured values
typically vary between −50 and 50 MPa. The average stress values are *14 MPa.
Again, no differences were found between the stress states using different
quenching temperatures.
Figure 5 summarizes the measured stress values of the rods subjected to water
cooling from the stress relaxation temperature. Again, three quenching temperatures
were applied: (a) 860 °C, (b) 880 °C and (b) 900 °C. Average stress values are
inserted. For all three rods, notable compressive stress was induced due to the water
cooling. The average stress values are around −350 MPa. The notable role of the
quenching temperature was not found once again.

www.TechnicalBooksPDF.com
64 J. Majtényi et al.

Fig. 5 Residual stress

distributions of the examined
rack bar rods with
intentionally induced stress.
Quenching temperatures are
(a) 860 °C (b) 880 °C
(c) 900 °C

3.2 Stress Asymmetry

Figure 6 shows the stress asymmetry of the rods quenched from (a) 880 °C and
(b) 900 °C. The average values of stress asymmetry are inserted. In Fig. 6 b, there
is a high stress asymmetry value, being above 140 MPa. However, all other values
and the average values are low, being *40 MPa. Evaluating a detailed stress and
stress asymmetry mapping connected with macroscopic deformation examinations
in Ref. [7] revealed that solitary, outlier stress and/or stress asymmetry values are

www.TechnicalBooksPDF.com
Comparative Destructive and Non-Destructive … 65

Fig. 6 Stress asymmetry

distributions of the examined
rack bar rods quenched from
(a) 880 °C (b) 900 °C

not responsible for the bending tendency of the rods. Based on that conclusion, the
sole high stress asymmetry value found here is considered as an outlier data.
Figure 7 shows the stress asymmetry of the annealed rods with (a) 860 °C
(b) 880 °C and (c) 900 °C quenching temperatures. The average values of stress
asymmetry are inserted. The measured stress asymmetry values and average stress
asymmetry values are low, being around 20 MPa.
Figure 8 shows the stress asymmetry of the rods induced during direct water
cooling from the stress relaxation temperature. For all three rods, the stress
asymmetry values are low, the average values are *25 MPa.
Table 1 summarizes the measured eccentricity values of the rods before and after
machining and the calculated Δ values. The average stress values are also included.
The states of the rods are marked as follows: quenched: Q, quenched + tempered:
QT, intentionally stressed after stress relaxation: IS. It can be seen that in the
quenched and annealed states, lower stress values, stress asymmetry values and Δ
were measured. However, when the rods were subjected to direct water cooling
after stress relaxation, lower stress asymmetry, but high stress and Δ values were
measured. In this study, the high Δ values, that is, bending tendency of the rods
cannot be associated with high Δ values, as was seen in our previous study [7]. On
the other hand, the strong bending tendency is clearly associated with higher stress

www.TechnicalBooksPDF.com
66 J. Majtényi et al.

Fig. 7 Stress asymmetry

distributions of the examined
rack bar rods after annealing.
Quenching temperatures are
(a) 860 °C (b) 880 °C
(c) 900 °C

values within the rods. This correlation can easily be understood if one takes into
account that during macroscopic deformation examination, the material removal is
asymmetric. If low stresses are present within the rod, removing material from one
part of the rod will not affect the stress state notably and the rod will not tend to
bend. On the other hand, if the rod has a symmetric, high stress state, the asym-
metric material removal destroys the stress balance and the rod bends. This can also
easily occur during the tooth machining of the steering rack bar. The applied
quenching temperatures had no effect on the stress states and bending tendency of
the rods.

www.TechnicalBooksPDF.com
Comparative Destructive and Non-Destructive … 67

Fig. 8 Stress asymmetry

distributions of the examined
rack bar rods with
intentionally induced stress.
Quenching temperatures are
(a) 860 °C (b) 880 °C
(c) 900 °C

Table 1 Measured eccentricity Δ, average stress values and average stress asymmetry values of
the examined rods. quenched: Q, quenched + tempered: QT, intentionally stressed: IS
State TQ Eccentricity Eccentricity Δ Average Average
(°C) prior to after (mm) stress stress
machining machining (MPa) asymmetry
(mm) (mm) (MPa)
Q 880 1.040 1.525 0.485 −36 42
Q 900 0.250 0.510 0.260 −42 42
QT 860 0.520 0.330 0.190 −14 18
QT 880 0.440 0.515 0.075 −15 20
QT 900 0.350 0.335 0.015 −10 17
IS 860 0.045 −2.720 2.765 −344 29
IS 900 0.085 −3.075 3.160 −370 18
IS 900 0.060 −2.760 2.820 −340 25

www.TechnicalBooksPDF.com
68 J. Majtényi et al.

4 Summary

In the present manuscript, the correlation was searched for between the stress states
and bending tendency of steering rack bar semi-products. The steering rack bars
were examined in quenched, tempered, and intentionally stressed states. The effect
of quenching temperature was also investigated. The stress states were examined
using non-destructive (sampling-free) X-ray diffraction method, while bending
tendency was examined with macroscopic deformation, during which, material was
removed from one side of the rods. It was found that symmetric, weak stress states
formed after quenching and tempering and no bending were measured. However,
when the rods were intentionally stressed by direct water cooling after stress
relaxation heat treatment, symmetric, but large (*−350 MPa) stresses formed,
which was associated with a strong bending tendency. The results clearly showed
that in a high stress state, even if the stress state is symmetric, asymmetric material
removal leads to a strong bending tendency. The examined quenching temperatures
(860, 880, 900 °C) had no effect on the results.

Acknowledgements The described article was carried out as part of the EFOP-3.6.1-16-00011
“Younger and Renewing University—Innovative Knowledge City—institutional development of
the University of Miskolc aiming at intelligent specialisation” project implemented in the
framework of the Szechenyi 2020 program. The realization of this project is supported by the
European Union, co-financed by the European Social Fund.

References

1. Totten G, Howes M, Inoue T (2002) Handbook of residual stress and deformation of steel.
ASM International, Ohio
2. Krawitz AD (2001) Introduction to diffraction in materials science and engineering, Columbia
3. Krauss G (1980) Principles of heat treatment of steel. ASM ASM 15:240–245
4. Richard E (2006) Haimbaugh: Induction heat treating. The Materials Information Society, Ohio
5. Rudnev V (2003) Handbook of induction heating, Libary of Congress Cataloging—
in-Publication Data, USA
6. Xstress 3000 G3/G3R system Operating instructions and instrument documents, 2012
7. Majtenyi J, Benke M, Mertinger V, Kazinczi T. The effect of quenching temperature and
polishing force on the residual stress and deformation of rack bar semi-products, Mat Sci
Forum, in press

www.TechnicalBooksPDF.com
Dynamical Modelling of Vehicle’s
Maneuvering

Ákos Cservenák and Tamás Szabó

Abstract This chapter deals with a dynamical modelling of autonomous vehicle,

including different manoeuvrings. Solution of the state-space form was performed
using the Runge-Kutta method. A special purpose program has been written in the
Scilab software system. Manoeuvrings of different cornering and overtaking are
simulated.

Keywords Mechatronics
Autonomous vehicle
Runge-Kutta method

Cornering
Mobile robot

1 Introduction

Nowadays prevention of accidents is very important in modern vehicles. An

autonomous vehicle can sense faster the environment, than a human driver.
Furthermore, the modern vehicle can prevent more accidents, than an older one
[1–4]. The research and education of such problems inevitable are important. The
vehicle engineering education has been started in this semester at the University of
Miskolc. The preparation of the lecture notes is in progress. This paper is a con-
tribution to this task.
The autonomous vehicle can also be considered as a mobile robot. The motion
along a straight line can be easily performed, but the vehicle often has to move
along arch, when cornering or overtaking are taking place. However, the motion
along arch may have a more critical situation. When the vehicle moves along an
arch with high speed, the vehicle might suffer instability which is called under
steering or over steering behaviour [5].
The linear single track model is used to describe motions of the vehicle, which
are detailed in [5]. The solution of a differential equation system is carried out by

Á. Cservenák (&)
T. Szabó

Robert Bosch Department of Mechatronics, University of Miskolc,
Miskolc, Hungary
e-mail: cservenak.akos@uni-miskolc.hu

© Springer International Publishing AG 2017 69

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_7

www.TechnicalBooksPDF.com
70 Á. Cservenák and T. Szabó

Runge-Kutta numerical method written in the Scilab software system. The program
can compute vehicle lateral slip angle, relative and absolute yaw angle, yaw rate
and the track of vehicle’s center of gravity in time.
Examples of different manoeuvrings, including lane changing, overtaking,
cornering in different cases will be shown for a vehicle [6] at constant velocity.
These examples are very useful in engineering education.

2 Dynamics of Vehicle’s Movement

A single track model is shown in Fig. 1, which can describe the dynamics of a
four-wheel vehicle. The notations are given as follows:
• av and ah : lateral slip angles of front and rear tire, respectively,
• b: vehicle slip angle,
• d: steering angle,
• wV : vehicle yaw angle,

Fig. 1 Linear single track xv

vehicle model with required δ
angles

αv
yE V

ψv

xE
β

yv S
ψv

αh

www.TechnicalBooksPDF.com
Dynamical Modelling of Vehicle’s Maneuvering 71

• w_ V : vehicle yaw rate,

• V point: the center of front wheel,
• H point: the center of rear wheel,
• S point: the center of gravity
• xv and yv coordinate axes attached to the vehicle
• xE and yE coordinate axes of the reference system
In the ideal case non-holonomic constraints would determine the motion.
However, our model is more realistic, the velocity of the center of a wheel usually
does not lie in its plane. This is due to the fact that the friction is considered and
lateral slip always experienced in case of cornering. Therefore, this model can also
be applied to determine critical speed in different situations. To answer such
question it is necessary to solve the dynamics of the vehicle.

3 Equations of Dynamical Model

The formulation of the equations of the dynamical model can be found, e.g., in [5].
A short summary will be given in the sequel.
The self-steering gradient “EG” is an essential quantity to characterize the
behaviour of a moving vehicle. Its value remains constant in our investigations
during motion. It depends on the mass m, the wheelbase l, the distance between
front wheel and center of gravity lv , the distance between rear wheel and center of
gravity lh and the cornering stiffnesses ca;h and ca;v . Its value can be determined by
the following formula:


m lh
ca;h  lv
ca;v
EG ¼
: ð1Þ
l ca;v
ca;v

When the value of “EG” is lower or higher than zero, the behavior of vehicle is
called over steering or under steering, respectively.
For a given path the cornering radii are known at each point. To follow this path
the steering angle is needed, which can be determined by the following formula:

1  
d¼
l þ EG
v2 ; ð2Þ
q

where v is the velocity of the centre of gravity. Otherwise, if the steering angle is
given, one can determine the cornering radius as follows:

l þ EG
v2
q¼ : ð3Þ
d

www.TechnicalBooksPDF.com
72 Á. Cservenák and T. Szabó

The equation of the motion is given in state-space form:

x_ ¼ A
x þ b
u; ð4Þ

where the “A” matrix is the system matrix

2 3
a11 a12 0
A ¼ 4 a21 a22 0 5; ð5Þ
1 0 0

and “a11 ”, “a12 ”, “a21 ” and “a22 ” are given as:

1 ca;v
l2v þ ca;h
l2h

a11 ¼ 
; ð6Þ
v h
ca;v
lv  ca;h
lh
a12 ¼  ; ð7Þ
h
1 ca;v
lv  ca;h
lh
a21 ¼ 1 
; ð8Þ
v2 m
1 ca;v þ ca;h
a22 ¼ 
: ð9Þ
v m

h is the moment of inertia and the

2 state
3 vector
2 3is
x1 w_ v
x ¼ 4 x2 5 ¼ 4 b 5; ð10Þ
x3 wv

the dot denotes the derivation with respect to time, b is the control vector
2 ca;v
lv
3
h
b¼ 4 1
ca;v 5; ð11Þ
v m
0

finally, the system input is the steering angle, i.e., u ¼ d.

4 Simulation Results

In this Section the dynamical simulations of a Suzuki car will be carried out. The
vehicle dimensions [6] and the rest of the estimated parameters are listed in Table 1.
Three different manoeuvres, i.e., an overtaking, a turning to left direction have
been analysed in the Scilab software system.

www.TechnicalBooksPDF.com
Dynamical Modelling of Vehicle’s Maneuvering 73

Table 1 The parameters of the vehicle

Name of dimension Dimension Quantity Unit
Mass of vehicle m 775 kg
Moment of inertia h 2000 kg
m2
Front length lv 1.032 m
Rear length lh 1.343 m
Wheel base l 2.375 m
Front cornering stiffness ca;v 60000 N
rad
Rear cornering stiffness ca;h 45000 N
rad

This car is regarded as a supermini car since its weight is relatively low and very
popular to use within cities. In the following examples the maximum velocity of the
car will be 70 km/h.

4.1 Overtaking

Firstly, an overtaking of another vehicle has been computed at constant velocity

v = 50 km/h. The manoeuvre of the vehicle is controlled by given steering angle.
The steering angle d ¼ 0 along a straight line (before t1 ¼ 3s; between t2 and t3 ;
and after t4 ). During the first phase of overtaking the steering angle d ¼
d0
sinðx
ðt  t1 ÞÞ is changing between t1 and t2 , where d0 ¼ 2 and x ¼ 2:45 rads ,
while in the second phase the steering angle d ¼ d0
sinðx
ðt  t3 ÞÞ is changing
between t3 and t4 . Runge-Kutta method is used for the solution of the differential
equation system.
The results, i.e., yaw angle, vehicle yaw rate, vehicle lateral slip angle and the
steering angle as input are shown in Fig. 2.
The track of the vehicle and the curve of the center of gravity are also obtained in
awareness of the velocity and the time function of absolute yaw angle. The track is
illustrated in Fig. 3, while the curve of the performed path is in Fig. 4.

4.2 Turning in Left Direction

A turning of vehicle in left direction has been described in the sequel. The velocity
is kept constant, i.e. v = 15 km/h. In this example a car will take a left turn. The
cornering radius is prescribed q ¼ 4:5 m, and the car runs along an arch of a quarter
of a circle. While getting close to the corner (t1 \2s) and leaving (t [ t2  3:7s) it
the steering angle should be zero to drive the car along straight lines. To start the
cornering the steering angle is changed abruptly to a constant value according to

www.TechnicalBooksPDF.com
74 Á. Cservenák and T. Szabó

Fig. 2 The slip-, the yaw-, the steering angles and yaw rate of movement during overtaking

Fig. 3 The track of vehicle during overtaking

Fig. 4 The curve of path during overtaking

www.TechnicalBooksPDF.com
Dynamical Modelling of Vehicle’s Maneuvering 75

(2) and the time of manoeuvre is calculated with the help of the velocity and the arc
length. Then at the end of the cornering the steering angle is changed back to zero.
The results for the vehicle yaw angle, the vehicle yaw rate, the vehicle lateral slip
angle and the steering angle are shown in Fig. 5. One can see that the trajectory of
the steering control has got sharp changes, however the response curves of the
vehicle, i.e., the vehicle yaw rate and the vehicle lateral slip angle are smoother. The
track of the vehicle and the curve of path are illustrated in Fig. 6.

Fig. 5 The slip-, the yaw-, the steering angles and yaw rate of movement during left turning

Fig. 6 The track of vehicle and curve of path during left turning

www.TechnicalBooksPDF.com
76 Á. Cservenák and T. Szabó

Fig. 7 The track of the vehicle turning at 70 km/h velocity

Repeating the computation at velocity 70 km/h but with the same steering
control approach the result is shown in Fig. 7. It can be seen that the vehicle cannot
keep the desired lane, which may lead to accident. It shows, the advantage of a
simulation program one can perform numerical experiments to determine the
appropriate steering controls for different velocities, which can be implemented into
an autonomous vehicle or mobile robot.

5 Conclusions

Dynamical modelling of a linear single track model has been analyzed numerically
in this paper. The vehicle can be controlled by prescribing the steering angle or the
cornering radius. The solution of the model provides the vehicle yaw angle, the
vehicle yaw rate, the vehicle lateral slip angle, the steering angle and the path.
Two manoeuvring problems have been examined. In the first one steering
control was applied for an overtaking at velocity 50 km/h. In the second problem
the cornering radius was given and the turnings have been performed at 15 and
70 km/h. In the latter case instability of the manoeuvre was experienced.
The two examples demonstrate the efficiency of the applied single track model.

www.TechnicalBooksPDF.com
Dynamical Modelling of Vehicle’s Maneuvering 77

Acknowledgements This research was carried out in the framework of the Center of Excellence
of Mechatronics and Logistics at University of Miskolc.

References

1. VDA, Verband der Automobilindustrie e.V (2015) Automation from driver assistance systems
to automated driving
2. Meiyuan Zhao (2015) Advanced driver assistant system, threats, requirements, security
solutions, Intel Labs
3. Gietelink O, Ploeg J, De Schutter B, Verhaegen M (2006) Development of advanced driver
assistance systems with vehicle hardware-in-the-loop simulations. Veh Syst Dyn 44
(7):569–590
4. Golias J, Yannis G, Antoniou C (2002) Classification of driver-assistance systems according to
their impact on road safety and traffic efficiency. Transp Rev 22(2):179–196
5. Schramm D, Hiller M, Bardini R (2014) Vehicle dynamics, modeling and simulation, Springer.
ISBN 3-5403- 6044-1
6. Suzuki Swift 1.0 GLX technical specifications, http://www.carfolio.com/ specifications/
models/car/?car = 85745

www.TechnicalBooksPDF.com
Developing a Rotary Internal Combustion
Engine Characterised by High Speed
Operation

László Dudás

Abstract The paper deals with the development of a new internal combustion
engine having a rotary piston. The introduction presents the short history of the
evolution of the rotary combustion engines. The main part of the paper introduces
the new patented internal combustion engine structure that has three rotational parts
only: the rotary piston—rotor—, the rotary housing—rotary chamber—and the
synchronizing gears. After the description of the structure, the work of the engine is
discussed, compared to the usual internal combustion engines and the advantages
and disadvantages are analyzed. Those properties that make possible the high speed
operation are emphasized. Besides the constructional characteristics some manu-
facturing tasks are also presented, especially the very important precision finishing
manufacturing of the working surfaces of the rotor and the rotary chamber.

1 Introduction

Though the development of automobile engines focuses on the electric motors

special areas may use the new, efficient versions of the conventional internal
combustion engines. Such area may be the motors of light sport airplanes and large
drones for carrying humans, where the use of the patented construction of the
author’s rotary engine is promising. Using the new industrial ceramics as motor
material adumbrates high efficiency of engines working quasi adiabatic manner at
high temperature. The paper introduces a new rotary internal combustion engine
construction that characterized by extremely high rotational speed ability owing to
its simplicity and frictionless working.
There is a long evolutionary history of rotary type engines that resulted in the
new motor presented in this paper. Two parallel ways of the inventions led to the

L. Dudás (&)
University of Miskolc, Miskolc, Hungary
e-mail: iitdl@uni-miskolc.hu

© Springer International Publishing AG 2017 79

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_8

www.TechnicalBooksPDF.com
80 L. Dudás

Fig. 1 Perna’s invention is a

modified screw compressor
that works as an engine [11 ]

mentioned invention. The one is the screw compressor line and the other is the
progressive cavity pump line. The screw compressor line started with the patent of
Krigar in Germany [1] and continued with different innovative types having
changing pitch, improved profiles and angle variation between the axes of the rotors
[2–7]. In the last years appeared Perna’s construction that can be used as internal
combustion engine as it has compressing and expanding sections as shown in
Fig. 1.
The progressive cavity pump line started with the patented construction of
Moineau in United States in 1932 [8]. His construction characterized by a rotor that
moved with a planetary motion in the static chamber substituting the rotations of the
rotor and the chamber. The special helicoid surfaces of the rotor and chamber
formed closed cavities that moved along the axis conveying fluid. Other patented
versions apply fixed axes for the rotor and the rotary chamber with eccentricity [9,
10]. The version that is close to the discussed motor invention has two regions of
the threads with different pitches. This construction can perform compression or
expansion, depending on the rotational direction [9]. Figure 2 shows an example.
Among the above-mentioned constructions the suggestion of Perna as the
member of the screw compressor type family is suitable for internal combustion
engine [11]. Similarly, applying compressing and expanding sections for the pro-
gressive cavity type machines a motor can be constructed as shown in Fig. 3. This
patented new motor [12] is the subject of the following discussions. The con-
struction is characterized by continuously changing pitch and an elliptical rotor
section curve. In the following the next sections will describe the construction, the
advantageous and disadvantageous features. Among the advantages the properties
that make especially suitable the construction for working with high rotary speed,
will be emphasized. The perfect work of the construction depends on the small gaps
between the rotor and the rotary chamber, so a section will analyze the grinding
possibility of the rotor.

www.TechnicalBooksPDF.com
Developing a Rotary Internal Combustion Engine … 81

Fig. 2 Murrow–Giffin type turbine [9]

Fig. 3 The patented engine construction [12]

2 The Suggested New Internal Combustion Engine

The new engine practically is a spatial Wankel engine [13]. According to the
original idea of Wankel, the triangle-like rotor and the chamber also rotated around
fixed axes. This construction was modified later by Paschke [14] to provide an easy
input of fuel-air mixture and exhausting of the hot expanded gas. Paschke dismissed
some disadvantage of the motor, but its advantage, the clear rotary motions, was

www.TechnicalBooksPDF.com
82 L. Dudás

lost. Our patented engine construction unites the Wankel principle of clear rota-
tional motion with the easy sucking in the fuel and the easy exhausting, as can be
followed in Fig. 3.

2.1 The Build-up of the Motor

This motor is one of the simplest internal combustion engine constructions in the
world. It has only three working parts: the rotor 1, the rotary chamber 2 and the
synchronizing gear 9, similarly to the original Wankel-idea. The rotor and the rotary
chamber have helical working surfaces that close cavities. These surfaces are
conjugate surfaces in the relative motion. A very small gap between the surfaces
provides the frictionless working. There is an e eccentricity between the two axes.
The intersection area along the axes is constant, equal to the difference in the area of
close curves 18 and 19. The cavities are larger at the two ends of the engine and
become smaller in the middle section because of the continuously changing pitch.
There are synchronizing gears 7, 8 and 9 to provide the 1:2 ratio between the rotor
and rotary chamber. A spark plug 17 is applied in the small cavity section.

2.2 The Working of the Motor

When the rotor and the rotary chamber rotate the closed cavities move from the left
intake side to the middle of the motor. The opened cavity at the left side of the
motor sucks in the fuel-air mixture. Then the cavity closes and becomes smaller and
smaller and compresses the mixture. In the middle the spark plug starts it up. The
hot gas expands the cavity and such a manner rotates the rotor and the rotary
chamber. Finally the cavity opens at the right hand side of the motor and the
expanded gas effuses from the cavity and has been removed totally by the moving
walls of the cavity. The gears synchronize the rotation of the chamber and the rotor
and the axle 9 serves as the output shaft. This shaft can also be used for starting the
motor.

3 Positive and Negative Properties of the Engine

In this section the advantageous properties and the disadvantages will be discussed,
in many times in comparison to the well-known Otto motor.
The main advantage of this motor is its simplicity. It results in many consecutive
advantages.

www.TechnicalBooksPDF.com
Developing a Rotary Internal Combustion Engine … 83

3.1 Cost

The simple construction results in low production and maintenance cost, long
lifespan and increased reliability. The motor works without valves and valve syn-
chronizing mechanisms, has less parts, less mass, less volume requirement, needs
less material than the usual Otto motors. The knowledge of setting the suitable
synchronizing of the valves for Otto motors is large, many books deal with it in
detailed form. The cooling of the exhaust valves is especially hard problem in Otto
motors. The calculation of the perfect length of the exhaust pipes and position of the
inlet and exhaust manifolds is a sort of art.

3.2 Efficiency

In this motor there is no waste energy because of valves, springs, alternating mass
of piston, friction between cylinders and pistons and between cylinders and piston
rings, between valves at valve-seats, between the valve stems and valve controlling
mechanisms. The frictionless working of the rotor without any sliding seal in the
chamber allows very high rotational speed in opposite to the Wankel engine too.
The high rotational speed means high power. The function curve of the compres-
sion and expansion can be set freely choosing proper pitch-function along the
length of the motor. There is no possibility for interfusion of the content of the
different cavities, the cavity contents are separated in every moment of the working,
the gas moves linearly with the cavities. The required turbulence of the fuel vapor
to achieve speedy combustion can be generated in the suitable form of the intake
slot. The air pollution can be limited owing to prolonged and repeated sparking.
There is no need for exhauster and sound damper as the pressure of the exhaust gas
can be as low as in the environment. Without water cooling the temperature can be
25–35% higher [15, 16]. Using industrial ceramics for the rotor and the rotary
chamber the cooling can be avoided and very efficient quasi adiabatic working
cycle can be realized owing to the extreme high combustion temperature, see
Table 1. The combustion temperature can be higher—1350 °C—than the melting
point of the metals. Using glass ceramics having low thermal conductivity the quasi
adiabatic process can be achieved, a low heat rejection engine can be made and the

Table 1 Engine material properties [17]

Material Thermal conductivity Max design temperature
(Wm−1K−1) (°C)
Metallic 54 1000
High performance 43 1350
ceramics
Glass ceramics 4 1000

www.TechnicalBooksPDF.com
84 L. Dudás

high temperature results in higher efficiency and less not burnt material in the
exhaust gas.

3.3 Noise

There is no noise resultant from valves and valve controlling mechanisms and
wearing in valves. There is no vibration, the motor rotary parts can be balanced
statically and dynamically. This is the main dynamical advantage, the rotary parts
simple rotate. The form of the combustion chamber decreases the effect of the
periodicity of explosions, the axial forces have a small acceleration effect at the
beginning of the combustion, the tangential forces that cause the rotation act
through longer space and time interval, causing smoother operation.

3.4 Properties Resulting in High Speed Operation

Owing to frictionless working of rotor and rotary chamber the speed is not limited
by the friction, lubrication. There are no alternating, accelerated parts in the motor.
Moreover, choosing suitable function for pitches in the intake section a continuous,
constant intake speed can be achieved. As the red line [18], so the physical
RPM-limiting factor is very high for this construction, it can easily outperform the
5,000–7,000 RPM of gasoline cars or even the 9,400 RPM of Mazda RX-8 Wankel
engine. It seems to be not a problem to exceed the very high 19,000 RPM redline
value of race motorcycle engines or 20,000 RPM of modern formula one cars. The
rotation speed of internal combustion engines increases when the size decreases.
The top speed of the smallest reciprocating engine of the world is 30,000 RPM
[19]. Such small size engine applies special fuel and ignition by compression. For
the suggested motor construction the RPM is limited by the rolling bearings or by
the centrifugal forces arising in the rotary chamber. Gears can work up to
40,000 RPM [20].

3.5 Disadvantages

Unfortunately, there are some disadvantages:

• Complex surface of the rotor and rotary chamber need special technology
• The small gap between rotor and rotary chamber needs special attention in the
timing of the explosion
• The inertia of the oil sealing in the small gap between chamber and rotor
surfaces decreases the efficiency

www.TechnicalBooksPDF.com
Developing a Rotary Internal Combustion Engine … 85

• Higher noise caused by rolling bearings opposite to sliding bearings of Otto

motors.
To avoid friction a small gap is required between the rotary chamber and the
rotor, similar to screw compressors. To achieve good sealing a very small gap is
essential, so special surface finishing method needed. This technology will be
presented and analyzed in Sect. 4. Unfortunately, this technology increases the cost
of manufacturing.

4 Manufacturing Considerations

The manufacturing considerations of this new motor type were presented in [21] in
detailed form. The manufacturing of the parts of the motor is easy except the rotor
and the rotary chamber. These parts have special helical surfaces having changing
pitch, so the usual methods applied for producing threads and worms can’t be
applied. The chamber is produced with uniting two halves of it. The inner helicoids
of the chamber machined in the halves by milling and the grinding is also possible
with the small diameter disk form grinding wheel when the two halves are
assembled. But the efficiency of such a process is very low.
In case of rotor the outer helical surface can be manufactured more easily. The
machining can be imagined using a Gellért-type polygon turning machine tool, or
milling is also possible with disk form tools. The finishing needs grinding and for
this the process and special grinding machine and grinding wheel that was invented
by the author can be used. This latter technology will be discussed in the following
subsection.

4.1 Analyzing the Grinding Possibilities

This section deals with the analysis of grinding of the rotor. The need of grinding is
justified by the required small gaps to achieve perfect sealing between the cavities.
The problem is similar to the sealing of screw compressor rotors [22, 23]. As the
rotor has a special helicoidal surface with changing pitch, conventional grinding
using a surface of revolution shape grinding wheel is impossible. The problem is
similar to grinding of tapered worms or hourglass worms. This problem is solved
by the author who proposed a new grinding machine structure and special grinding
wheel [24]. The patented grinding machine shown in Fig. 4 is suitable for grinding
quasi-helical surfaces, e.g. tapered and hourglass worms. The novelty is that the
machine applies a 1:1 rotation ratio between the workpiece and the grinding wheel,
so the two surfaces are conjugate surfaces. For the generation of the grinding wheel
surface an enveloping by the rotor surface method was applied. Such process can be
performed using the Surface Constructor software application developed by the

www.TechnicalBooksPDF.com
86 L. Dudás

Fig. 4 Grinding machine for

grinding quasi-helical
surfaces

Fig. 5 Designing and checking the surfaces in the Surface Constructor

author. If the grinding wheel can be generated without undercutting, i.e. with clear
enveloping, then the rotor can be grinded exactly with the generated grinding
wheel. In the followings this generating process and the checking against undercut
will be presented.
Figure 5 shows the Surface Constructor software application. The software was
applied to create the rotor surface, for generation of the inner surface of the rotary
chamber and for generation of the grinding wheel surface for the rotor grinding. The
upper left window shows the rotor, the upper middle window shows the chamber

www.TechnicalBooksPDF.com
Developing a Rotary Internal Combustion Engine … 87

Fig. 6 Checking against undercut using Fi2 functions

surface and the upper right window shows the grinding wheel surface. The lower
windows show the grinding wheel and the rotor in contact, and some other win-
dows show the undercut analysis. These results appeared after inputting the
required expressions and data (after entering the Fi2 grinding machine kinematics
and setting the required Zeta2, Rho2 and Tau2 expressions and parameters for the
generating process). To avoid undercut it was important to determine the right
tilting angle for the rotary axis A of the grinding machine. Earlier experiences
adumbrated that for the modelling of tapered or hourglass worm the best grinding
tilting angle is equal to the least pitch angle. This value was 5° for the helicoids of
the rotor.
The checking against undercutting was accomplished using the special visual-
ization capability of Surface Constructor. In a dedicated window we can scan the
whole Tau2, Zeta2 parameter domain of the F22 grinding wheel surface and the
window for every point draws the Rho2 = Rho2(Fi2) function. Such a function can
reveal all types of local undercut and global cut with its shape. The scientific
background that includes the theoretical basis of the Surface Constructor can be
found in [25]. If the Rho2 = Rho2(Fi2) function shows inflection or a local max-
imum, then the inspected point of F22 is a local undercut point. The perfect grinding
wheel points are characterized by a smooth valley form. If all the F22 points in the
analyzed Tau2, Zeta2 parameter domain show perfect shape, then the surface has
not undercuts or edges. The mapping window acts as a scanning window at the
same time, so gives an easy way to scan all the points of the F22 grinding wheel

www.TechnicalBooksPDF.com
88 L. Dudás

surface. By scanning the points the problematic locations can be detected. The
checking against undercut proved that even a 0° tilting angle results in perfect
grinding surface. Figure 6 shows some points with the Rho2 = Rho2(Fi2) function
curves demonstrating the method. The grinding wheel can be produced for exact
grinding of the rotor surface, but making such a wheel is a complicated, expensive
process, thus it is suggested in case of mass production of rotors.

5 Summary

The paper introduced a new, patented rotary internal combustion engine. The
advantages and disadvantages were discussed and the suitability for high speed
operation was justified. Finally the finishing operation of the rotor using a special
form of grinding was presented. In this process the special capability of Surface
Constructor for detection undercuts was applied. In the future the analysis of the
perfect pitch functions and the proper sparking will be analyzed. The mathematical
modelling of the pressures and forces will be the following task to determine the
moment and the power of the motor.

Acknowledgements This research was partially carried out in the framework of the Center of
Excellence of Mechatronics and Logistics at the University of Miskolc. The financial support is
acknowledged.

References

1. Cicerale FH (2009) Kenne Bell Supercharger Ford Mustang Install—Superbad. Mustang360°,

article published 1 March 2009. http://www.mustangandfords.com/how-to/engine/mmfp-
0903-2004-ford-mustang-kenne-bell-supercharger-install/. Accessed 1 Sept 2016
2. Abe YT (1996) Fluid rotary apparatus having tapered rotors. United States patent US
5533887 A, 1996
3. Becher U (2003) Twin screw rotors and displacement machines containing the same. United
States patent US 20030152475 A1 2003
4. Jacobsson BO, Gabelli A (1990) Screw compressor. European patent EP0376373 A1, 1990
5. Kapp B. Improvements in screw pumps. Patent GB696732 (A), 1953
6. Martin H, Sachse R (2006) Rotors for a screw compressor. Patent EP 1722104 A2, 2006
7. North MH (2001) Screw vacuum pump. United States patent US 20010024620 A1, 2001
8. Moineau LRJ (1932) Gear Mechanism. United States patent US 1892217 A, 1932
9. Murrow KD, Giffin RG (2009) Axial flow positive displacement turbine. United States patent
2009/0226336 A1, 2009
10. Wiedenhoefer JFA (2009) Non-contact seal for positive displacement capture device. United
States patent US 20090220369 A1, 2009
11. Perna V (2003) Equipment with mutually interacting spiral teeth. United States patent. US
2003001267 A1, 2003
12. Dudás L (2015) Rotary piston internal combustion engine. Hungarian patent HU230082,
2015

www.TechnicalBooksPDF.com
Developing a Rotary Internal Combustion Engine … 89

13. NSU Wankel rotary engines and cars. http://www.tripod.com/rotary/pg05.htm. Accessed 2

Sept 2016
14. Der Wankelmotor und sein Erfinder. http://www.nsuro80.ch/das-auto/der-motor/. Accessed
10 Sept 2016
15. 2012 Ceramic Rotary Engines, Inc: Ceramic Rotary Engine, http://www.youtube.com/watch?
v=YnfFEf7wyww, Elérve: 10 Sept 2016
16. Stosic N, Smith IK, Kovacevic A (2003) Opportunities for innovation with screw
compressors. In: Proceedings of the institution of mechanical engineers, Part E: J Pro
Mechanical Eng 217(3):157–170. http://www.staff.city.ac.uk/*ra601/oportsvi.pdf. Accessed
23 Aug 2016
17. Salazar F (1998) Internal combustion engines, Department of Aerospace and Mechanical
Engineering. University of Notre Dame, Notre Dame, IN 46556, 30 April 1998
18. Wikipedia. “Redline”, http://en.wikipedia.org/wiki/Redline. Accessed Sept. 21, 2016
19. Ronney PD (2016) Hydrocarbon-fueled internal combustion engines: “the worst form of
vehicle propulsion… except for all the other forms”, Online publication: http://ronney.usc.
edu/whyicengines/WhyICEngines.pdf. Accessed 11 Sept 2016
20. Triveni Engineering & Industries Ltd. product catalogue: High Speed Gears and Gearboxes.
(2016) Online publication: http://www.trivenigroup.com/download/gearbox-catalogue.pdf.
Accessed 11 Sept 2016
21. Dudás L, Biró M, Novák LL (2016) Construction modeling and manufacturing analysis of a
new rotary combustion engine. In: INES 2016 20th Jubilee IEEE international conference on
intelligent engineering systems, June 30-July 2, 2016, Budapest, Hungary pp 89–94
22. Holmes CS (2004) Inspection of screw compressor rotors for the prediction of performance,
reliability, and noise. In: International Compressor Engineering Conference, Paper 1692.
http://docs.lib.purdue.edu/cgi/viewcontent.cgi?article=2691&context=icec. Accessed 3 Sept
2016
23. Kacalak W, Majewski M, Budiak Z (2015) Worm gear drives with adjustable backlash.
J Mech Rob 8(1):7–17
24. Dudás L (1992) Grinding machine, for grinding non-surface of revolution surfaces, especially
conical and globoid worms. Hungarian patent HU P9003803, 1992
25. Dudás L (2010) Modelling and simulation of a novel worm gear drive having point-like
contact. In: Horváth I, Mandorli F, Rusák Z (Eds.) Proceedings of TMCE 2010 Symposium,
Ancona, pp 685–698

www.TechnicalBooksPDF.com
Simulation Methods in the Vehicle Noise,
Vibration and Harshness (NVH)

Károly Jálics

Abstract The chapter introduces the simulation methods (MBS, FEM, SEA)
which are generally used in the vehicle NVH. Alongside also hybrid methods will
be introduced. An overview will be given about the usage of the methods
depending on the frequency range for the simulation and prediction of the NVH
behaviour of full vehicle and its components.

1 Introduction

The basis of the simulation methods was laid down already in the sixties of the past
century (e.g. Finite Element Method, Statistical Energy Analysis). The widespread
usage of them was blocked through the primitive computer technology. Through
the fast development of the computers also numerical methods in general walked
through an enormous evolution.
Vehicle NVH became more and more important in the past decades, since the
regulations concerning environmental protection (pass-by noise) and also vehicle
comfort expectations became more severe. Not only the production quality, the
perfect material selection in the passenger compartment, etc., but also the acoustic
impression became an important criterion of the quality of a vehicle. Also the low
interior noise level, excellent speech intelligibility, excellent sounding of the HiFi
system and the specific sound design of a vehicle became more and more important.
Also the NVH experts have recognized soon that simulation methods enhance the
development process of a vehicle. The goal of the simulations is the calculation,
respectively the prediction of the NVH behaviour of a full vehicle or its compo-
nents. The selection of the proper simulation method is based on the investigated
frequency range, one certain method cannot used for the full acoustic frequency

K. Jálics (&)
University of Miskolc, Miskolc, Hungary
e-mail: machijk@uni-miskolc.hu

© Springer International Publishing AG 2017 91

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_9

www.TechnicalBooksPDF.com
92 K. Jálics

Fig. 1 Methods of the

vehicle NVH simulation
depending on the complexity,
boundary conditions a
frequency range [1]

range (0–20 kHz). Therefore 4–5 methods have to be involved successively, or the
usage of the co-simulation has to be considered in order to cover the full domain.
However, in general the simulation methods are currently not able to predict the
perceived noise of the passengers even for a single operating condition.
A general brake down of the simulation methods used for vehicle NVH simu-
lation depending on the frequency range of interest, complexity and system
dimensions is shown in Fig. 1.
As an example for serial usage of different the NVH calculation methods of a rail
vehicle can be given. In that case the rail—wheel interaction is calculated with
MBS. The obtained forces are serving as excitations for the FEM calculation of the
bogie. The vibration of the bogie exciting the vehicle body and the internal pas-
senger cavity. The sound pressure level can be finally calculated by the SEA model.
The main goal of this chapter is to represent the utilization of this method for NVH
applications. Also the advantages and disadvantages of the individual methods will
be pointed out.

2 Simulation Methods

2.1 General Aspects of a Simulation Method

In general the main task the simulation is the conversion of a real object, a complex
system or a physical problem into a simplified mechanical/mathematical

www.TechnicalBooksPDF.com
Simulation Methods in the Vehicle Noise, Vibration … 93

substitution model. Thereby the following terms should be fulfilled, which is

basically a balance act between model accuracy and prediction quality:
• idealized assumption
• as accurate as possible
• as simple as possible.
The range of applications of the methods is of course limited through the sim-
plification. As the next step mathematical equations will be set up. By discrete
masses, damper and stiffness finite number (e.g. a few hundred) of differential
equations or by continuous masses, damping and stiffness theoretically infinite
number (e.g. in the practice even some million) of partial differential equations.
Depending on the described aspects, the following methods can be used for the
NVH simulation.

2.2 Multi-Body Simulation

As a non-continuous method for low frequency calculations the Multi-Body

Simulation (MBS) has been established. MBS is generally used for the calculation
of the dynamic system performance in the automotive engineering. The field of
applications is vehicle dynamic, comfort (NVH), powertrain and engine design. The
method uses typically discrete masses (rigid or flexible bodies) which are connected
each other or to the surrounding with linear or non-linear coupling elements (spring
or damper) or with kinematic constrains. The bodies may undergo large transla-
tional and rotational displacements. The bodies have max. 6 independent degrees of
freedom which can be limited through boundary conditions or constraints. The
displacement conditions are described with generalized coordinates with the
number of the degrees of freedom. Mathematically described the system contains
regular 2nd order differential equations as seen bellow:

M€ _ þ KqðtÞ ¼ FðtÞ
qðtÞ þ DqðtÞ ð1Þ

where M, D, K are time invariant matrices (n x n) for mass, damping and stiffness,
q = (q1; q2; … qn) time dependent general coordinates and F(t) includes the
excitation forces. By large systems or by higher frequencies the assumption of rigid
bodies is no more acceptable. In that case the Finite Element Method (FEM) can be
used.

2.3 Finite Element Method (FEM)

FEM is well known since several decades, and widely used in the engineering
sciences. The FEM delivers reliable absolute results in the buckling, strength, heat

www.TechnicalBooksPDF.com
94 K. Jálics

transfer and fatigue analysis. FEM is also a state of the art method of the vehicle
NVH simulation. It also provides reliable results concerning natural frequencies,
transfer functions for a single part or for a not too complex system as well as in the
higher frequency range (*1000 Hz). However, by a complex vehicle FEM model,
which contains a structure (body, chassis, etc.) and also air cavities (passenger
compartment, trunk, etc.) a coupled simulation (acoustic-structure) calculation is
needed. In that case the frequency range is limited to a few hundred Hz (<300 Hz),
since the complexity of the model is already enormous, the model can have several
million degrees of freedom, which makes the computational effort expensive. In this
chapter only the NVH aspects will be pointed out.
The simulation process starts with the division of the body into many small
elements (ashlar, tetrahedron etc.). The discretization is very important at the NVH
calculations [2]. The element length determines basically the upper frequency limit
of the calculation. Equivalent to the Nyquist sampling theorem, every mode shape
(waves) should be sampled with a sufficient number of degrees of freedom (or
elements). The maximum element length depending on the requested frequency (f),
the bending wave propagation speed (cB) and N number of elements per wavelength
is given as follows for structures:

kB cB
dmax ¼ ¼ ð2Þ
N Nf B

In the practice at least 6 elements per wavelength should be used which delivers
a good agreement between accuracy and computation time. An example is shown in
Fig. 2 for a steel plate with 2 mm thickness. If e.g. a calculation up to 500 Hz
should be performed, with N = 6 and 10 elements per wavelength the maximum
element length are 33 mm respectively 20 mm. The continuity conditions at the

Fig. 2 Maximum element length depending on the frequency (for steel plate with 2 mm
thickness, by 6 and 10 elements per wavelength)
Simulation Methods in the Vehicle Noise, Vibration … 95

bounding surfaces of the surface or volume elements lead to a linear equation

system, similar to the Eq. (1) in the previous chapter. The solution of the
homogenous equation system delivers the natural frequencies and the mode shapes.

2.4 Statistical Energy Analysis (SEA)

Through the statistical averaging of the modes the problem can be solved simpler.
For that case the Statistical Energy analysis (SEA) delivers good results for the
higher frequency range. SEA was developed already in the sixties of the last
century for the calculation of dynamic behaviour of complex structures at high
frequencies. Compared to deterministic methods (e.g. FEM) SEA derives average
energetic values which means a major limitation of the method. As the method’s
name indicates, SEA provides a statistical framework and does not give the exact
deterministic solution. In SEA 2 types of averaging will be considered:
• Frequency averaging: averaging of the modes, velocities, powers, etc. in a
certain frequency band, general in third octave band
• Spatial averaging: averaging over the points of observation and excitation. All
phase relevant information will be lost.
The mathematical principles of SEA can be illustrated as shown in Fig. 3. A system
with 2 subsystems is given by the following power equilibrium equations [3, 4]:

Pi þ Pji ¼ Pij þ PL;i ð3Þ

Pj þ Pij ¼ Pji þ PL;j ð4Þ

where Pi and Pj are the input power to the subsystems, PL,i, PL,j power losses
(internal) and finally Pij and Pji are the transmitted power from one subsystem to the

Fig. 3 Typical frequency ranges in the vehicle NVH calculations

96 K. Jálics

other. Equation 2 can be written also in a generalized matrix form with the intro-
duction of the modal density Ni:
2 3 2 32 3
Pi gi N i    gni Nn Ei =Ni
6 .. 7 6 .. .. .. 76 .. 7
4 . 5 ¼ x4 . . . 54 . 5 ð5Þ
Pn gin Ni    gn N n En =Nn

where [Pi] is the vector of the input powers; [E/N] is the vector of the modal
energies of the subsystems; ηn are the internal loss factor (ILF), ηnn are the coupling
loss factors (CLF) and Ni is the modal density. For most systems, especially large
systems, many subsystems are not connected. This means that the matrix will
contain many zero off-diagonal elements and therefore a sparse matrix. The input
powers are assumed to be known. If the modal densities of the subsystems, the ILF
and the CLF, are known, the total energies of subsystems can be determined. The
ILFs are experimentally identified in most cases or extracted from a database. CLF
also can be determined experimentally.
This method is mathematically easy and fast compared with FEM. There is a
matrix to be solved and the elements will be directly related to the number of the
subsystems. In this method, the number of modes per frequency band (or modal
density) acts the major role and the accuracy of obtained results largely depends on
it. Regarding the minimum required number of resonant frequencies in a certain
frequency band, different authors suggested different values. In general several
modes per frequency band deliver accurate results, however, with only one mode
per band accurate results can be achieved, if the other interacting subsystems have a
sufficient number of modes ( 3) in the same frequency band.

3 Hybrid Methods

For the low frequency range, the calculation of the modes is for example done using
the Finite Element Method. In the high frequency domain, where high modal
overlap occurs, SEA is better suited as mentioned in the introduction. A predicting
gap exists in the mid frequency range. Furthermore, the structure-borne transmis-
sions are not well predicted. To cover the mid frequency gap, e.g. a hybrid
FEM/SEA method can be applied. (Also other combinations of methods, e.g.
FEM/MBS, FEM/BEM, etc. existing.) The subsystem with the low modal density is
modelled e.g. with FEM. The results will be coupled to the SEA calculation. An
example is given in Fig. 4, where a vehicle firewall is shown for a transmission loss
analysis. The exact shape of the firewall is built up in FEM, which is fully coupled
to neighbouring chambers modelled with SEA. For this test the average SPLs of
both chambers can be simply calculated in SEA.
Simulation Methods in the Vehicle Noise, Vibration … 97

Fig. 4 Hybrid FEM/SEA

model of a vehicle firewall
for transmission loss
calculations [5]

4 Summary

In the past chapters simulation methods were introduced, which are widely used in
the vehicle NVH simulation. FEM is one of the most powerful and all-round
method also in the vehicle NVH. Although there is a frequency limitation by
approx. 300 Hz for the full vehicle coupled acoustic-structure simulation. Also the
computational efforts (millions of DOFs) are quite high. SEA is mathematically
easy and fast in comparison with FEM. One of the main advantages of SEA is that
it can help to identify the major contributor in the overall energy of the system. Also
it is easy to monitor the effects of changes in the design. The main disadvantages of
the method are the loss of phase information, the results are valid for a subsystem
and not for a certain point and low frequency problems because of the low modal
density at low frequencies of the structures. MBS is widely used in the low fre-
quency range, but the implementation of elastic parts condensed with the FEM can
enhance the frequency range.

References

1. Zeller P (2012) Handbuch Fahrzeugakustik, Springer Verlag

2. Knothe K, Wessels H (2008) Finite Elemente—Einführung für Ingenieure, 4th edn. Springer
Verlag, Berlin, Heidelberg
3. Lyon RH (1975) Statistical energy analysis of dynamical systems. MIT Press, Cambridge,
Massachusetts
4. Lyon RH, Dejong RG (1995) Theory and application of statistical energy analysis, 2nd edn.
Butterworth-Heinemann, Boston
5. https://www.esi-group.com/software-solutions/virtual-performance/va-one/va-one-modules/
hybrid-module
Optimal Damping of Random Excited
Systems

Ferenc Knopp

Abstract Through the case study of a simple mechanical system (mass, springs,
damper) we analyse the effect of the damper coefficients on the output variance. The
system described by a continuous differential equation is substituted with a discrete
time (sampled) autoregressive moving average ARMA (3, 2) time series model.
The displacement input (road profile) also characterized by an ARMA (p, q) sta-
tionary stochastic process. A variance transformation formula is derived which uses
the (simulated) discrete impulse response function and the autocorrelations of the
input signal. This formula is applied to the mechanical system searching the optimal
damping which gives the minimal output variance.

1 Signal Transmission of Time Series

A discrete time Auto-Regressive Moving Average ARMAðn; mÞ time series model

is defined below. The “a-members” at time k look back to the previous values of the
yk output signal and weigh them, while the “b-members” do the same with the uk
input signal.

yk ¼ ða1 yk1 þ . . . þ an ykn Þ þ ðb0 uk þ b1 uk1 þ . . . þ bm ukm Þ ð1Þ

Its stability condition and other features can be found in the literature [2, 5, 8, 9].
In our technical case study we shall see that a continuous ordinary differential
equation also gives an ARMA model, if we apply discrete numerical approxima-
tions for the derivatives.
The output is the wk impulse response function (weighing sequence) if u0 ¼ 1
and u1 ¼ 0; u2 ¼ 0; . . . namely the input is the discrete unit impulse. Every input

F. Knopp (&)
Budapest University of Technology and Economics (BME), Budapest, Hungary
e-mail: knopp@mogi.bme.hu

© Springer International Publishing AG 2017 99

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_10
100 F. Knopp

value starts an uk -stretched impulse response and the superposition of them at a

time k builds up the yk output signal value.

X
k X
k
yk ¼ wk u0 þ wk1 u1 þ . . . þ w0 uk ¼ wki ui ¼ wi uki ð2Þ
i¼0 i¼0

This is the so called convolution in which the direction of summation is

changeable, and extendable to the infinite past. In the special case if the input is the
uk ¼ 1 unit jump than we get the vk step response function which is the sum of the
impulse response function till the actual time k.

X
1 X
k
yk ¼ wi uki ðconvolutionÞ vk ¼ wki  1 ðstep responseÞ ð3Þ
i¼0 i¼0

In simple cases the impulse response function can be expressed by an explicit

formula, but generally is numerically created “whirling” the model and storing the
values in a vector large enough. For example at the yk ¼ a1 yk1 þ b0 uk ARMA (1,
0) system wk ¼ ak1 b0 ðk ¼ 0; 1; 2 ; . . .Þ.

2 Transformation of Variances

Assuming stationary random signals with zero mean values: M½ui  ¼ 0, M½yi  ¼ 0.
Then the l-distance auto-covariance and the normalized auto-correlation for the
input series:

Rl ¼ Ruu ðlÞ ¼ M½ui uil  Ruu ð0Þ ¼ D2u ruu ðlÞ ¼ Ruu ðlÞ=D2u ð4Þ

Writing the output signal—created with infinite convolution—into the first row
and first column of a table, multiplying and summing the table, we just get the
square of it. (Assuming the existence of the limit)

y k2 w0u k w1u k 1 w2u k 2 w3u k 3 ...

w0u k w02 u k u k w0 w1 u k u k 1 w0 w2 u k u k 2 w0 w3 u k u k 3 ...
w1u k 1 w0 w1 u k 1u k w12 u k 1u k 1 w1w2 u k 1u k 2 w1w3 u k 1u k 3 ...
w2u k 2 w0 w2 u k 2u k w1w2 u k 2u k 1 w22 u k 2u k 2 w2 w3 u k 2u k 3 ...
w3u k 3 w0 w3 u k 3u k w1w3 u k 3u k 1 w2 w3 u k 3u k 2 w32 u k 3u k 3 ...
... ... ... ... ... ...
Optimal Damping of Random Excited Systems 101

Applying the main value operator on the cells of the table, the auto-covariances
of the shifted signals appear. M½uki ukj  ¼ Ruu ði  jÞ ¼ Rij .

D y2
w02 R0 w0 w1 R1 w0 w2 R2 w0 w3 R3 ...
w0 w1 R1 w12 R0 w1w2 R1 w1w3 R2 ...
w0 w2 R2 w1w2 R1 w22 R0 w2 w3 R1 ...
w0 w3 R3 w1w3 R2 w2 w3 R1 w32 R0 ...
... ... ... ... ...

Summing this infinite table, we get the output variance D2y . From the main
diagonal R0 ¼ D2u can be taken out, and in the first two neighbour sub-diagonal R1
is common, in the second two R2 , and so on… Taking out D2u the general variance-
transformation formula is ready. In the case of independent input (white noise)
only the first sum remains, summing the main diagonal. (See also [4].)
" ! #
X
1 X
1 X
1
D2y ¼ D2u  w2i þ2  wi wi þ l  ruu ðlÞ ð5Þ
i¼0 l¼1 i¼0

Practically speaking a finite 50  50 table—assumed to be large enough—was

used, because the impulse response function of a stable system goes to zero fast.
A dynamically increasing table size is applicable with checking the relative
accuracy.

3 Dynamic Models of a Mechanical System

The system in Fig. 1 can be seen as a (rotated) vehicle suspension system which is
excited by the uðtÞ road displacement as input signal. Applying the NEWTON’s II.
law of motion [1, 6, 7] for the mass and the weightless node, we get a system of
second order differential equations for the yðtÞ and zðtÞ output signals:

m  €y ¼ c2  ðz  yÞ þ d  ð_z  y_ Þ
ð6Þ
0  €z ¼ c2  ðy  zÞ þ d  ð_y  z_ Þ þ c1  ðu  zÞ

Ordering these left the uðtÞ input signal appears on the right hand side.

m  €y þ d  y_ þ c2  y  d  z_  c2  z ¼ 0
ð7Þ
 d  y_  c2  y þ d  z_ þ ðc1 þ c2 Þ  z ¼ c1  u
102 F. Knopp

Fig. 1 A mechanical c2 [N/m]

m [kg]
vibrating system c1 [N/m]

d [N/(m/s)]

u(t) [m] z(t) [m] y(t) [m]

After Laplace transformation [3], there exists a system of algebraic equation for
the transformed functions. From here the yðsÞ global output can be expressed easily
with the help of the Cramer’s rule.

m s 2 d s c2 ( d s c2 ) y(s) 0
u (s) (8)
( d s c2 ) d s (c1 c2 ) z ( s ) c1
0 (d s c2 )
c1 (d s c2 ) N (s)
y(s) u (s) u (s) (9)
2 D( s)
m s d s c2 (d s c2 )
(d s c2 ) d s (c1 c2 )

The denominator of the transfer function is the DðsÞ characteristic polynomial.

The coefficients of it are the output (left) side coefficients of the continuous dif-
ferential equation. Similarly NðsÞ shows the structure of the input (right) side.
After simplifications we find a proportional-differentiating (PD) system with
third order delay (T3) after all our system is a (PD-T3) type.

v þ mðc þ c Þ  €
md  yðtÞ yðtÞ þ c1 d  y_ ðtÞ þ c1 c2  yðtÞ ¼
1 2
ð10Þ
_
¼ c1 c2  uðtÞ þ c1 d  uðtÞ

The discrete numerical approximation over a D ¼ Dt width four member grid

yk3 ; yk2 ; yk1 ; yk at its (virtual) centre:
yk1  yk2 uk1  uk2
y_ ð Þ ffi _ Þffi
uð ð11Þ
D D
yk  yk1  yk2 þ yk3
€yð Þ ffi ð12Þ
2  D2

vÞffi yk  3yk1 þ 3yk2  yk3

yð ð13Þ
D3
Optimal Damping of Random Excited Systems 103

Substituting these into the (10) continuous differential equation and grouping the
coefficients of the grid values:
 
md mðc1 þ c2 Þ
yk  þ þ
D3 2  D2
 
3md mðc1 þ c2 Þ c1 d c1 c2
þ yk1   3  þ þ
D 2  D2 D 2
 
3md mðc1 þ c2 Þ c1 d c1 c2
þ yk2    þ
D3 2  D2 D 2
  ð14Þ
md mðc1 þ c2 Þ
þ yk3   3 þ ¼
D 2  D2
 
c1 c2 c1 d
¼ uk1  þ þ
2 D
 
c 1 c2 c1 d
þ uk2  
2 D

Expressing yk the derived discrete ARMA (3, 2) time series model:

yk ¼ ða1 yk1 þ a2 yk2 þ a3 yk3 Þ þ ðb1 uk1 þ b2 uk2 Þ ð15Þ

4 Seeking the Optimal Damping

Figure 2 shows the running result of a QUICK BASIC computer program (under
DOS, see Appendix). There are no comments on it, but the subroutine (SUB) names
help to associate them to the functions and formulas. In the Visual Basic environment,
these can be inserted as modules (without any graphical statement). They can be
realized in other high level programming language too (C/C++ etc.).
Choosing the d ½N/(m/s) damping as free parameter we get a curve for the
output variance with a minimum (Fig. 3). The smallest value 1:85 [ 1 so at these
mechanical and road features the output variance always exceeds the input variance.
At zero damping the two springs (c1, c2) are in serial switch (softer system), but if
the damping goes to infinity (rigid rod) the first spring (c1) works only (harder
system).
Our purpose was to explain the theoretical basis of the problem. Further exe-
cuted and desired investigations are not described here. The important question is
the characterization of the road profile and connections to some “stochastic
resonance”.
104 F. Knopp

Fig. 2 Simulation of the discretized mechanical vibrating system. Step response (green):
m ¼ 1000; c1 ¼ 10000; c2 ¼ 10000; d ¼ 3000; D ¼ 0:2, Input signal (red): yk ¼ 0:5  yk1 þ
0:2  yk1 þ 2:65  uk

Fig. 3 Output variance in

function of the damping
Optimal Damping of Random Excited Systems 105

Appendix
106 F. Knopp
Optimal Damping of Random Excited Systems 107

References

1. Bokor J, Nándori E, Várlaki P (2000) Studies in vehicle engineering and transportation science.
Hungarian Academy of Sciences and BME, Budapest (Hungary)
2. Box EPG, Gwilym MJ Time series analysis forecasting and control, Holden-Day
3. Fodor G (1967) Lineáris rendszerek analizise. Műszaki Könyvkiadó, Analyse of Linear
Systems (in Hungarian)
4. Lantos B (2001) Theory and design of control systems I. Akadémiai Kiadó, (in Hungarian)
5. Michelberger P, Szeidl L, Várlaki P (2001) Applied process statistics and time series analysis.
Typotex Kiadó, (in Hungarian)
6. Sályi B, Michelberger P, Sályi I (1991) Kinematics and kinetics Tankönyvkiadó, (in
Hungarian)
7. Thomson TW (1972) Theory of vibration with applications. Prentice-Hall, Inc.
8. Tusnádi G, Ziermann M (1986) Time series analyses. Műszaki Könyvkiadó, (in Hungarian)
9. Várlaki P (1986) Introduction to the statistical system identification. Műszaki Kiadó, (in
Hungarian)
Application of Knowledge-Based Design
in Computer Aided Product Development

György Hegedűs

Abstract In the recent years, it can be observed that the products are launched by
rapidity and short time. The lifecycles of user equipment has been shortened, the
flow of information accelerated, the freely available information more widely
accessible. One of the conditions of competitiveness for the arisen problems is the
rapid and optimal solutions. Using a software or system, which supports the
knowledge based design (KBD) the time and costs of the product development
phase can be reduced. In this chapter the results of a product design and devel-
opment process implemented in a PLM system will be presented on a ball screw
drive mechanisms focusing on its returning guide.

Keywords Knowledge-Based design (KBD)
Product development
Computer

aided design (CAD)

1 Introduction

In the last decades the computer aided design was the answer for quicker solutions.
These systems and software supported the geometric modelling and manufacturing
planning, especially (CAD/CAM). Due to the development the integrated systems
were launched, which included the different simulation and analysing capabilities of
the products. Nowadays the application of product lifecycle management
(PLM) systems is an essential part of the product development. One of the
advantages of these systems that knowledge-based design supporting modules are
available, it is possible to create user-defined macros and algorithms, furthermore
decision making and selection criteria arise during the design can be defined pre-
viously. Taking advantage of these properties the time of product design and
development can be reduced significantly.

G. Hegedűs (&)
University of Miskolc, Miskolc, Hungary
e-mail: hegedus.gyorgy@uni-miskolc.hu

© Springer International Publishing AG 2017 109

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_11
110 G. Hegedűs

2 Brief Overview of Knowledge-Based Design

Knowledge-based design focuses on the product design, development and its

related procedures [1]. In a KBD environment the design process can be accelerated
by reusing previously defined techniques, methods and procedures and these can be
integrated into the desired tasks or workflows which are involved in the product
development processes. The flexibility and complexity of knowledge-based design
environment largely depend on the reused geometrical data, previously defined
expressions and algorithms (see Fig. 1).
Formerly these environments included non-parametrised geometries (for
example: neutral *.stp, *.iges, or converted CAD files). Complex CAD software
support the parametrised surface, solid and assembly models, thus the knowledge
based design and development can be achieved by the reused parametrised
geometry. In product data and product lifecycle management software the mathe-
matical expressions and logical relations or functions extend the capability of the
knowledge-based environment. Most CAD, PDM and PLM software support user
defined applications are developed in different programming languages such as
C/C++, C#, Visual Basic, Java and specific built-in programming language [3].
Users can create their own applications that automatically perform repetitive tasks,
accelerate design procedures, and automatically generate complex geometries. In
such a case the user defined software obtains the parameters, equations, logical
relations and functions from the CAD system and evaluates the numerical com-
putations, algorithms end geometry creations. After that the evaluated parameters
will be set into the CAD system. The flexibility, interactivity and time saving of
repetitive tasks are the advantages of these KBD applications, however the disad-
vantages of these KBD environments are that all of the computations, numerical
algorithms and communications between the CAD system and user defined soft-
ware have to be programmed [4–7].

Fig. 1 Flexibility and complexity of a KBD application [2]

Application of Knowledge-Based Design in Computer Aided … 111

Knowledge-based design methods and tools can include rigid or variable

geometry data, the integration of calculation and simulation procedures into the
design process, or the application of problem-oriented software solutions that can
be integrated into the design environment. Although one main advantage of
knowledge-based design is that it makes existing, proven solutions available for
specific tasks, this also involves some drawbacks that must be acknowledged [2, 8].

3 Application of Ball Screw Drives in Automotive

Equipment

Gothic-arc profile ball screw motion transforming mechanisms are widely used to
transform the rotational movement into linear movement and vice versa. Due to the
automotive development these motion transformation mechanisms have been
appeared in the electrically operated steering systems.
The electrically operated steering systems (Electric Power Steering, EPS) have
introduced in passenger car steering systems during the last years. The use of these
systems was originally limited to smaller vehicles, because the power density of the
electronic parts and the energy available from the on-board wiring was not sufficient
to serve bigger vehicles and higher steering powers.
New technologies enable the general use of EPS in the superclass now. The
advantage of electric power-assisted steering compared to hydraulic power steering
is that it is activated only when needed (energy is fed only when the car is steered).
Due to this the energy waste and CO2 emission are less.
The paraxial drive (EPSapa) is characterized by low system friction and high
efficiency. The possible customer applications range from dynamic sports cars and
upper mid-size cars to high load vehicles such as off-road vehicles and vans. Due to
the combination of ball screw mechanism and toothed belt drive with paraxial drive
is ideally suited for the customer’s differing performance requirements. The wide
range of positioning possibilities of the servo unit allows optimum use of the
installation space on the vehicle. The linear movement of the steering rack is
generated from the rotational movement of the electric motor combined with a
toothed belt drive and a ball screw mechanism (see Fig. 2). The ball screw
mechanism generates low noise and rigid connection between the steering gear and
the vehicle body is possible [10].
Another solution of the application of ball screw drive mechanisms in EPS
systems is the so-called rack-concentric steering system (EPSrc). The toothed belt
drive is eliminated and the ball nut is connected with a hollow shaft and directly
driven by an electric motor (see Fig. 3). The concentric arrangement of the hollow
shaft requires a special servo motor, because the rack of the steering passes through
the motor. The main advantage of the EPSrc system is its compactness, the dis-
advantage that due to the missing gear multiplication of the toothed belt drive the
electric motor of an EPSrc has to produce a twice as high torque at the same output
power level.
112 G. Hegedűs

Fig. 2 Electric power steering with paraxial drive (EPSapa) [9]

Fig. 3 Electric power steering with rack concentric drive (EPSrc) [11]

4 Application of Ball Screw Drives in Automotive

Equipment

The customer’s demand requires the development of ball screw drive mechanisms.
Gothic-arc profile ball screw motion transforming mechanisms are commonly used
in machine tools and the demand for high-lead ball screws is increasing due to
high-speed manufacturing. The one of the problem of the high-lead ball screw drive
is its manufacturing (determination of the profile of form grinding tool, collision
avoidance of the quill and ball nut during manufacturing). The lead of ball screw
mechanisms in an EPS system has a typical value of 5–10 mm. On low-lead ball
Application of Knowledge-Based Design in Computer Aided … 113

Table 1 The main design Parameter Description Unit

parameters of ball nut
P Pitch of ball screw drives mm
D3 Internal diameter of ball nut body mm
Ln Length of thread mm
Dpw Pitch circle diameter mm
Rpr Radius of profile arc mm
c Centre shift of gothic arcs mm
Dw Ball diameter mm

Fig. 4 Redesigned returning guides with KBD

screw drive mechanism the main problem is the arrangements of returning guides in
the ball nut. Therefore new returning guides were designed to create a
knowledge-based environment.
The KBD environment consists the reused design parameters of ball nut (see
Table 1), the logical rules and functions.
The results of the new returning guide design variants are shown in Fig. 4, where
the optimal solution was selected by value analysis.

5 Conclusions

This chapter has presented a product development process on the returning guide of
a ball screw drive mechanism, which would be applied in automotive steering
system. To reduce the time and cost of the development process a knowledge-based
design environment was developed in a PLM software. Due to the favourable
advantages of KBD system the time of the returning guide design variants gener-
ation is reduced and the selection of optimal solution can be carried out easier.
114 G. Hegedűs

Acknowledgements This research was carried out as part of the TÁMOP-4.2.1.B-10/2/

KONV-2010-0001 project with support by the European Union, co-financed by the European
Social Fund. This research was supported by the European Union and the State of Hungary,
co-financed by the European Social Fund in the framework of TÁMOP-4.2.4.A/2-11/1-2012-0001
‘National Excellence Program’.

References

1. Hirz M, Harrich A, Rossbacher P (2011) Advanced computer aided design methods for
integrated virtual product development processes. Comput-Aided Des Appl 8(6):901–913.
doi:10.3722/cadaps.2011.901-913
2. Hirz M, Dietrich W, Gfrerrer A, Lang J (2013) Integrated computer-aided design in
automotive development: development processes, geometric fundamentals, methods of CAD,
knowledge-based engineering data management. SpringerLink, Bücher. Springer Berlin
Heidelberg. ISBN:9783642119408
3. Rocca GL (2012) Knowledge based engineering: between AI and CAD. Review of a language
based technology to support engineering design. Adv Eng Inform 26(2):159–179.
ISSN:1474-0346. doi:http://dx.doi.org/10.1016/j.aei.2012.02.002
4. Tzivelekis CA, Yiotis LS, Fountas NA, Krimpenis AA (2015) Parametrically automated 3D
design and manufacturing for spiral-type free-form models in an interactive CAD/CAM
environment. Int J Interact Des Manufac (IJIDeM) 1–10. ISSN:1955-2505. doi:10.1007/
s12008-015-0261-8
5. Hegedűs Gy (2016) Newton’s method based collision avoidance in a CAD environment on
ball nut grinding. Int J Adv Manuf Technol 84(5):1219–1228. ISSN:1433-3015. doi:10.1007/
s00170-015-7796-5
6. Gomes S, Bluntzer JB, Mahdjoub M, Sagot JC (2007) Collaborative, functional and
knowledge based engineering using a PLM environment. In: DS 42: Proceedings of ICED
2007, the 16th international conference on engineering design, Paris, p 12
7. Catic A, Malmqvist J (2007) Towards integration of KBE and PLM. In: DS 42: Proceedings
of ICED 2007, the 16th international conference on engineering design, Paris, p 12
8. Jayakiran Reddy E, Sridhar CNV, Pandu Rangadu V (2015) Knowledge based engineering:
notion, approaches and future trends. Am J Intell Syst 5(1):1–17. doi:10.5923/j.ajis.
20150501.01
9. Lenksysteme ZF (2013) ZF servolectric—electric power steering system for passenger cars
and light commercial vehicles. Brochure
10. Harrer M, Pfeffer P (2017) Steering handbook. Springer International Publishing.
ISBN:978-3-319-05449-0. doi:10.1007/978-3-319-05449-0
11. Bochert A, Cäsar T, Eichler J (2010) Steering system. ATZextra worldwide 15(11):140–143.
ISSN:2195-1489. doi:10.1365/s40111-010-0252-5
Elementary Calculations for Deflection
of Circular Rings

Géza Németh

Abstract The traction drives requires initial tension. Let us imagine such kind of
epicyclic drive or harmonic drive. In both cases the annular wheel is made of spring
steel and its load is bending. The stiffness of the ring is large enough to produce the
required normal force between the annular and the planetary wheels or between the
ring and the flexible, cup shaped wheel. The transmission ratio depends on the ratio
of planetary and sun wheel in the planetary drive, or the difference between the
circumferences of ring wheel and the cup shaped wave wheel on the harmonic
drive. Both the deflection of ring wheel and the number of planetary wheels or the
number of deflection wave increase the load carrying capacity and decrease the
bending moment. The problem of deflection can be managed by differential
equation or any kind of energy method. The analysis of equivalence of the out-
comes closes the paper.

1 Introduction

The well-known epicyclic gear drive and harmonic gear drive are produced in large
series by many manufacturers. Their load carrying capacity, accuracy and the large
transmission ratios suggest their application fields. These drives are used mainly in
heavy drives, robots and vehicles, and also in kinematic drives. The latter appli-
cation field usually requires less torque capacity and less accuracy.
The kinematic drives can be epicyclic traction drives and also traction drives
with generated deflection waves that similar to the harmonic drives. The gear drives
are shape closing ones, and the traction drives are force closing. The gear drives do
not require initial tension and can transmit large tangential forces. It is true sig-
nificantly in the case of harmonic gear drives, where the number of teeth meshing
simultaneously is huge. It is worth to notice, that the amplitude of deflection waves

G. Németh (&)
University of Miskolc, Miskolc, Hungary
e-mail: machng@uni-miskolc.hu

© Springer International Publishing AG 2017 115

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_12
116 G. Németh

necessary for the pairs of gears to leave or enter the meshing is relatively large. In
case of traction drive there are no such constraints to increase the wave amplitude.
There are two distinct problems where the friction drive is analysed in this paper.
There is an epicyclic traction drive with one outer and one inner connection and
there is a harmonic traction drive. Both drives have a ring wheel with relatively
large stiffness and usually fixed to the stand. The ring wheel is loaded radially by
N number of radial forces, Fr and the same number of tangential forces Ft which are
less than the maximum frictional forces, Ft \l0 Fr , where l0 is the coefficient of
static friction between the ring wheel and the mating element N ¼ 2; 3; 4; . . .. The
mating elements are the planet wheels or the flexible cup shaped wheel, depending
on the type of drive. The number of planet wheels, but the number of waves
influences the transmission ratio. A friction drive can transmit sufficient load only
with large compressive forces or greater number of compressive forces.
Our purpose is to develop a kinematic traction drive with a relatively large
transmission ratio also for a relatively large torque. The compressive force between
the mating elements should be assured by the elastic deflection of the ring wheel.
Similar problem was analysed in our previous papers [1, 2]. The ring wheel was a
helical torsion spring there and it was analysed by a spatial model. When the
cross-section b  h; the mean diameter d and the material properties E and m of the
ring wheel is known, and also the number of mating planet wheels (in case of
epicyclic drive) or the number of waves (in case of harmonic drives), we can
estimate the necessary range of deflecting on.
Let us analyse a circular ring with rectangular cross-section, loaded equally with
an equally distributed arbitrary number of concentrated radial forces Fr , shown in
Fig. 1. Do find the relation between the force and the deflection. We suggest two
distinct solutions for this problem. Both are well known in the literature [3]. The
first one is the deflection curve for a bar with a circular centre line, the second one is
the deflection of the curved bar calculated by the Castigliano’s theorem. The rate of
thickness and radius of the ring is relatively small, so the ring is considered as a thin
one.

Fig. 1 Circular ring loaded

by equally distributed radial
forces
Elementary Calculations for Deflection of Circular Rings 117

Fig. 2 Models for deflection curve and for Castigliano’s theorem

It is also interesting to find relationship between the calculated results. The

equation of the deflection curve obtained from differential equation is available in a
polar coordinate system, and its origin is the centre of the circular centre line O. The
Castigliano’s theorem provides its result also in polar coordinate system, but the
problem is triple statically indeterminate, and due to symmetry we analyse only a
range of p=N arc. One end of the arc is built in, the other is free. In this model the
fixed end has no any deflection or slope of the deflection curve. Otherwise the
Castigliano’s theorem can be applied more conveniently, but its results are not in
the appropriate form, that is why the conversion is advisable. The two models are
shown by Fig. 2.

2 The Mechanical Model

2.1 The Bending Moment Along the Circular Centre Line

A thin closed bar with circular centre line is loaded by an N number of equally
distributed, equal concentrated forces, F acting radially outwards. The maximum
value of radial deflection is expected at the locations of forces and the maximum
negative radial deflection is expected in the middle of two equal forces. Due to
symmetry it is advisable to analyse a p=N arc instead of the full circle. Our
assumption is that the strain energy due to shear is negligible to that of the bending.
In the model, one end of the arc of the range of p=N angle is built in, its radius, r is
constant, the other end is free and loaded by a tangentially acting force,
F=½2 sinðp=NÞ and a concentrated bending moment, M0 . The arc coordinate, s ¼ ru
is started at the free end of the arc. The bending moment function along the arc is

Fr ð1  cos uÞ
M ¼ M0 þ
 : ð1Þ
2 sin Np
118 G. Németh

The free end has a deflection but it has no slope. Considering this constraint, the
bending moment, M0 is computable and the bending moment function (1) can be
expressed as
" #
Fr N cos u
M¼ 
 ; ð2Þ
2 p sin Np

and its value at u ¼ p=N is the maximum one,

  p 
Fr N
Mmax ¼  cot : ð3Þ
2 p N

2.2 The Deflection Curve

The differential equation of the deflection curve for a bar with a circular centre line,
considering the results of [1] and the bending moment function (2), where the radial
deflection, u is the wanted variable, is
" #
d2u Fr 3 cos u N
þu ¼
 : ð4Þ
du2 2IE sin Np p

Having solved the Eq. (4) we obtain the function of radial deflection as
" #
Fr3 N u sin u cos u  p  p 
uðuÞ ¼  þ
þ
 1 þ cot : ð5Þ
2IE p 2 sin Np 2 sin Np N N

Due to symmetries, it is sufficient to consider the range of arc coordinate as

uð0; p=NÞ. The positive values represent the radial increment of the points of bar
measured from the centre of the circular bar.

2.3 The Deflection by Strain Energy

The Castigliano’s theorem is suitable for the solution of statically indeterminable

problems of bars. Due to symmetry, only a part of the full circle should be con-
sidered, visualized by Fig. 3. The angle of arc is 2Nth part of the whole 2p.
To obtain the deflections at any point of the beam in the function of u, beside the
bending moment from the real loads (2), denoting by Mreal , The other part of the
imagined loads should also be considered. The sum of these components is
Elementary Calculations for Deflection of Circular Rings 119

Fig. 3 The cantilever beam

model for generating the
deflection and slope functions

M ¼ Mreal þ FPt mt þ FPr mr þ MP mw ; ð6Þ

where the so called relative bending moment functions are the partial derivatives of
the whole bending moment and their values are

mt ¼ r ½1  cosðu  uP Þ;
mr ¼ r sinðu  uP Þ; ð7Þ
mw ¼ 1:

Designating the generalized deflections by di and the relative bending moment

functions by mi , where i ¼ 1; 2; 3, the general formula for the deflections at any
point of uP is
p
ZN
r
di ð u P Þ ¼ Mreal ðuÞmi ðuÞdu: ð8Þ
IE
u¼uP

Having been carried out the integrations when one end of the beam at the angle
of p=N is built in, and the independent variable is u again, instead of uP , the radial
and tangential deflections are
  sin u
Fr3 1  p  Nh p i  p
dr ðuÞ ¼ sin u  1  cos u  u ;
2IE 2 N p N N 2 sinðp=NÞ
ð9Þ
   
p 
Fr 3 N 1 p N p  sin u u
dt ðuÞ ¼  u þ cot  sin u þ þ N cos u ;
2IE p 2 N p N sinðp=NÞ 2 sinðp=NÞ
ð10Þ
120 G. Németh

and the slope at any cross section is


Fr2 sin u N
dw ðuÞ ¼  u : ð11Þ
2IE sinðp=NÞ p

2.4 The Equivalence of the Formulae

Comparing the formulae (9) and (5) of the radial deflections, the differences
between them is obvious. And this is the expected result, too. The model of the
Castigliano’s theorem having a built in end and another constraint for the slope on
the free end, cannot supply the same values as the solution of the differential
equation. The equivalence of the formulae is supposed. It is important to prove this
fact. The task is interesting, because the handling of the problem is easier by
Castigliano’s theorem and on the other hand, the deflections can be more conve-
niently used, when the fixed point is the centre of the original circular bar.
Let us designate the latter deflections by uðuÞ and vðuÞ. There are three curves
of angle range p=N should be compared and located into the same coordinate
system, shown in Fig. 4. The first one is a circle with the centre point of O and
constant radius, r. The second one is the deflection curve with the same fixed point
and variable radius. The third one is the deflected cantilever of the built in circular
bar. Studying these curves the following geometrical formulae can be written for
the radial and tangential deflections.

p uðuÞ  dr ðuÞ
cosð  uÞ ¼ dt ð0Þ
; ð12Þ
N
sinðp=NÞ

p dt ðuÞ  vðuÞ
sinð  uÞ ¼ dt ð0Þ
; ð13Þ
N
sinðp=NÞ

where dt ð0Þ is the tangential deflection at the free end of the built in model that is
shown at Fig. 4. Substituting these transforming formulae in (9) and (10) the

Fig. 4 Deflections at the different coordinate systems

Elementary Calculations for Deflection of Circular Rings 121

deflections are available in polar coordinate system with starting point of the centre
of circular line, O.
( )
Fr3 N h p  p i cos u u sin u
uðuÞ ¼  þ 1 þ cot
þ
 ; ð14Þ
2IE p N N 2 sin Np 2 sin Np
( )
Fr3 N h p  p i sin u u cos u
vðuÞ ¼  uþ 1þ cot

 ; ð15Þ
2IE p 2N N sin Np 2 sin Np

Fr 2 sin u N
#ðuÞ ¼  u : ð16Þ
2IE sinðp=NÞ p

The formula (14) is really the same as formula (5) and the (15) also fulfil the
constraints and the experiences. The maximum value of the curvature, jmax is
expected at the location of the original force application, at u ¼ p=N.
Approximating the deflected curve as qðuÞ ¼ r þ uðuÞ, and considering the value
of the first derivative at that point to be zero, q0 ðp=N Þ ¼ 0, the equation of cur-
vature is fairy simple.



q  q00 q þ Np  Np a þ a Np cot2 Np
jmax ¼ ¼ 


 2 ; ð17Þ
q2 q  Np a þ a2 cot Np þ Np 1 þ cot2 Np

where a ¼ Fr 3 =ð2IEÞ: Equation (17) gives useful geometrical data, too.

3 Conclusion

The deflection of discrete points or the deflection curve for a bar with the circular
centre line, but for arbitrary number of loads are available in the literature. The
necessity of these equations is important to design a ring wheel for a simple
epicyclic traction drive or a harmonic traction drive. The analyses suggested the
equations in the convenient form. The formulae obtained from the model is suitable
for a preliminary calculation of some machine elements both in epicyclic and
harmonic traction drives.

Acknowledgements The research work presented in this paper/study/etc. based on the results
achieved within the TÁMOP-4.2.1.B-10/2/KONV-2010-0001 project and carried out as part of the
TÁMOP-4.1.1.C-12/1/KONV-2012-0002 project in the framework of the New Széchenyi Plan.
The realization of this project is supported by the European Union, and co-financed by the
European Social Fund. The research work was supported by the Hungarian Scientific Research
Found grants OTKA 29326 and Fund for the Development of Higher Education FKFP 8/2000
project.
122 G. Németh

References

1. Németh G, Németh N, Péter J (2015) Strength calculation of elastic elements in epicyclic

traction drive (in Hungarian). In: Sándor B (ed) Műszaki Tudomány az Észak-Kelet
Magyarországi Régióban 2015 Konferencia előadásai, Debreceni Akadémiai
Bizottság Műszaki Szakbizottsága, pp 213–219. ISBN:978-963-7064-32-6
2. Németh G, Péter J, Németh N (2013) A new type of epicyclic traction drive. Adv Mech Eng 1
(1):137–142. ISBN:978-963-473-623-3
3. Timoshenko S (1968) Strength of materials, elementary theory and problems. Princeton, New
Jersey, pp 442
Position- and Speed-Dependent,
Power-Absorbing Hydraulic Cylinder
with Mathematically Predictable
Characteristics

Lajos Albrecht, Ferenc Mészáros, Szilárd Szabó and Balázs Barna

Abstract Within the framework of the research a position- and speed-dependent,

power-absorbing hydraulic cylinder were constructed based on laboratory tests and
finite element numerical simulation studies. The elaborated cylinder construction
has a mathematically predictable characteristics and based on this feature can be
used for different applications in the field of kinetic energy absorbing.

Keywords Hydraulic power absorbing
Mathematically predictable characteris-

tics Articulation system

1 Introduction

Utilizing the funds won in tender no. KMOP-1.1.1-08/1-2008-0050, between 2009

and 2013, the associates of SOLID4D Kft. carried out the development of a
hydraulic cylinder that is suitable for application as a power-absorbing unit of
special characteristics depending on the displacement, with a very simple structure
[1]. Its special features and simple structure makes is possible to use this product in
a wide range of application areas; for example, in the machine and automotive

L. Albrecht (&)
Mechanical and Transport Engineering, University of Miskolc, Miskolc, Hungary
e-mail: lajos.albrecht@gmail.com
F. Mészáros
Industrial Engineering, University of Miskolc, Miskolc, Hungary
e-mail: ferenc.meszaros@innospectrum.hu
S. Szabó
Department of Fluid and Heat Engineering, University of Miskolc, Miskolc, Hungary
e-mail: aram2xsz@uni-miskolc.hu
B. Barna
Mechanical Engineering, Department of Machine Tools, University of Miskolc, Miskolc,
Hungary
e-mail: barna.balazs@uni-miskolc.hu

© Springer International Publishing AG 2017 123

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_13
124 L. Albrecht et al.

industry, too. In the last 3 years (2013–2016), progress was made in the field of
utilization concerning development and production, furthermore, a new type of
articulation systems for trailers and sliding articulated buses was tested.

2 Theoretical Bases

The clarification of theoretical questions and the mathematical formulation of the solution
took place as a solution for a problem occurred in connection with a research and
development topic. The basic idea was to utilize the dynamic pressure forming during the
movement of the piston in the hydraulic cylinders, and to make the dynamic run of the
pressure, occurring in the cylinder, predictable by regulating the outflow by a variable gap
in the cylinder. The preliminary mathematical analyses revealed the special behaviour
according to which, by applying a properly designed, changing throttle gap led by piston
displacement, a power-absorbing hydraulic cylinder (bumper, shock absorber) can be
created which automatically adapts to the load (depending on the speed and position of
the piston). The run of the braking force—as a function of the current speed of the piston
—changes automatically so that during its stroke length, the cylinder absorbs the kinetic
energy of the colliding object under all circumstances. The forces occurring during the
collision process are the smallest possible, i.e. it behaves ideally as a power-absorbing
bumper (shock absorber or damper). According to the proposed solution, longitudinal
grooves of varying cross section are prepared for the inner wall of the cylinder. Upon
loading (displacement of) the piston, the oil flows through these as throttles between the
cylinder wall and the piston. By properly creating the cross section of gaps and their axial
size distribution, the dynamic load caused by the force affecting the piston bar is of
predictable rate, and the displacement and its time course (function) can be determined in
advance according to the needs and adjusted to the given technical task.

3 Hydraulic Basic Research

As the first step, the numeric and laboratory test of the cross flow factor of gaps in
various shapes took place, in order to be able to determine the cross flow factor as a
function for hydraulic throttles of different cross section and length; under variable

Fig. 1 The apparatus serving for receiving resistance bodies with different throttles to be tested, in
an assembled status
Position- and Speed-Dependent, Power-Absorbing … 125

Fig. 2 Resistance bodies with boreholes of different shape (cushion) and the cross sections of
throttles

parameters. For these tests, a measurement tool was prepared with removable discs
(Fig. 1), to which we machined gaps of a length of 5, 10 and 15 mm with three
different cross sections (triangle, rectangle and circle) (Fig. 2).
The final purpose of the joint performance of the numeric tests and measure-
ments was to control all of the initial parameters for the further numeric simulations.
Thus, at the end of the validation process, it is possible to perform the preliminary
calculation of new variations by applying the calculations, avoiding the costly
manufacture and measurement of a high number of prototypes. The
three-dimensional hydrodynamic simulation calculations took place by applying the
Ansys Fluent program system [2].
The simulation calculations and the measurements were prepared with the
throttles of three different shapes, of lengths of 5, 10 and 15 mm but of the same
cross section (9 mm2). In Fig. 3, we present the measurement section established
for the reception of throttles, which contains the pressure tappings, as well—these
are located 50–50 mm before and after the throttles. In order to avoid cavitation, we

Fig. 3 The drawing of the measurement section containing the throttling element and the flow
picture obtained during the simulation calculations
126 L. Albrecht et al.

Fig. 4 Geometry of the calculation model

Fig. 5 The speed distributions emerging in the middle plane of boreholes in case of a pressure
difference of 60 bar

kept the exiting pressure at a level of pk  30 bar. Based on the features of the
university’s testing equipment, pressure differences between p  10  80 bar
could be established on the throttles. (Later on, we had the opportunity to perform
the control measurements also up to a pressure of 40 bar in the laboratory of a
company that deals with the manufacture of hydraulic parts).
The calculation performed with the Ansys-Fluent flow simulation program [2] took
place on the numeric model with the geometry drafted in Fig. 4, which fully corresponds
to the surfaces of the measurement section—drafted in Fig. 3—in contact with oil.
During the simulation calculations, we obtained very informative results, for
example, in the relatively small gaps, the speed distribution was calculated, as well;
showing the dead zones of the flow, as shown in Fig. 5.
The analyses were supported also by the determination of flow conditions
formed in the spaces before and after the gaps. The pictures of Fig. 6 show
examples of this.
We summarized the results obtained with the calculations and during the mea-
surements in Figs. 7 and 8, with comparative tests.
Figure 7 shows the pressure difference-volume flow diagram prepared based on
the measurement and calculation results. Based on the figure, it can be stated that
the measured Q(Δp) curves run very close to each other. A slightly higher volume
flow is detected only in case of the circle-shaped borehole. In Fig. 8 we can see the
Position- and Speed-Dependent, Power-Absorbing … 127

Fig. 6 Manifestation of speed distributions by trails

Fig. 7 The measured

pressure difference-volume
flow curves

Fig. 8 The measured

pressure difference-cross flow
factor curves
128 L. Albrecht et al.

dependence of the Kv cross flow factor from the pressure difference. In accordance with
Figs. 7 and 8 shows that in case of the rectangular and triangular borehole, the mea-
sured cross flow factors differ slightly from each other only in case of a small pressure
difference. In case of the triangle, here its value is smaller, which can be justified by the
corner design of the triangle—unfavourable from the aspect of the flow.
In case of the circular cross section, the cross flow factor moves decisively above
the tested range compared to the other cases, and the difference decreases with the
increase of the Δp pressure difference. The calculated cross flow factors are
4  14% smaller that the measured values. The bigger difference occurs at low
volume flows, and the difference gradually decreases with the increase of the
volume flow. The differences are not significant, probably during the manufacturing
procedures the dimensions are slightly different from the exact values applied at the
calculations. A further research task may be the revision and clarification of the
geometries applied at the measurement and the calculations and of other deter-
mining factors.
In Figs. 9 and 10, in case of a hydraulic cylinder equipped with a gap of
changing cross section, it shows the cross-flown quantities and the cross flow factor

Fig. 9 Correlation of the L 70

stroke and the Q flow rate in Dp=10 bar
case of various pressure 60
Dp=20 bar
differences
Dp=30 bar 50

Q [l/min]
Dp=50 bar 40

30

20

10

0
80 70 60 50 40 30 20 10 0
Abs(L) [mm]

Fig. 10 The change of the 0,7

cross-flow factor as a function
0,6
of the L stroke
0,5

Kv, max 0,4

Kv [-]

0,3 Dp=10 bar

0,2 Dp=20 bar

Dp=30 bar
0,1
Dp=50 bar
0,0
80 60 40 20 0 -20 -40 -60 -80
L [mm]
Position- and Speed-Dependent, Power-Absorbing … 129

Fig. 11 Draft of a measurement arrangement enabling variable piston position

as a function of the stroke (piston position) in case of various pressure differences.

These measurement results provided the basis for the establishment of the final
construction.
In order to be able to draw the diagrams of Figs. 9 and 10, we collected data by
using a measurement aid shown in the draft of Fig. 11 and the photo of Fig. 12,
furthermore, by measurement arrangement. It was possible to set the position of the
piston by the threaded spindle attached to the shaft. After carrying out the cross
flow test of the throttling gap of a cross section variable according to the given
function, established in the wall of the cylinder and of a length of 160 mm; it
became possible to draw the diagrams of Figs. 9 and 10 and to get a picture on the
evolution of the cross flow factor depending on the variable geometry of the
throttling gap.

Fig. 12 The measurement tool and measurement arrangement of cross flow tests
130 L. Albrecht et al.

The above described tests—at which the hydraulic simulation was performed by
the Department of Flow and Heat Engineering Machines of the University of
Miskolc, and the validation was conducted by the Department of Machine Tools on
a hydraulic load bench—were completed successfully. The aim of the study was to
get information on the dampening effect of the throttling gaps created in the wall of
the cylinder, the relationship between the gap size (shape) and the dampening
effect, i.e. the behaviour of the so called “cross-flow factor”—usually indicated as a
constant—as a function and on its values depending on the changing test param-
eters. The purpose of the joint performance of measurements and calculations was
to find out in what way and with what level of precision is it possible to estimate the
throttling effect with numerical simulation.

4 Testing of Prototypes and Steps of Utilization

By using the results of the research performed, several kinds of prototypes were
prepared and tested. In addition to the power absorbing hydraulic cylinders of
smaller diameter (28 and 50 mm) a complete articulation system was prepared for
buses, as well, which is able to replace or exchange those applied currently, with a
much simpler and more reliable design. Application for the industrial property right
protection has been submitted for this structure, as well. The successful testing of
the new-type articulation system took place in July this year (2016), built into a
435-type IKARUS articulated bus. Figure 13 illustrates its sketch. The tests per-
formed are detailed in a related article.

Fig. 13 Assembly draft of

the articulation system
Position- and Speed-Dependent, Power-Absorbing … 131

References

1. Bencs P, Barna B, Makó I, Szabó Sz (2011) Effect of gap geometry on flow losses. In:
Proceedings MicroCAD ’11 international computer science conference, Section D, Miskolc,
Hungary, pp 13–18
2. FLUENT: Fluent V6.2 User’s guide. Fluent Inc., Lebanon, New Hempshire, USA, 2005
 Part II
Technology
Utilization of the GD OES Depth Profiling
Technique in Automotive Parts Analysis

Tamás I. Török and Gábor Lassú

Abstract Vehicles often operate in rather harsh and even extreme environmental
conditions, so their many parts all should resist well against corrosion, wear and
other outer impacts, which surface phenomena need continuous developments and
frequent testing. One rather fast and quite effective surface analytical technique is
the Glow Discharge Optical Emission Spectrometry (GD OES), the applicability of
which is demonstrated here by its several laboratory uses to detect the elementary
composition in depth of some contaminated then plasma cleaned, pre-treated then
metal and/or organic coated (painted or varnished) or just properly surface modified
samples originating from different segments of automotive manufacturing or dis-
assembled cars put under maintenance and/or repair.

1 Introduction

There are quite many new and novel developments in surface engineering by means
of which one can enhance the surface properties (i.e. resistance to wear and cor-
rosion, decorative appearance, lubrication, solderability etc.) of several important
functional and structural parts of a vehicle. The two key methods of surface
engineering are surface coatings and surface modification as illustrated in Fig. 1.
In both surface treatment techniques the coated or modified outermost zones of
any component must be thoroughly controlled during their manufacturing via
non-destructive testing techniques, however, time to time their modified surface
properties must also be checked in depth through applying some additional and

T.I. Török (&)
G. Lassú

Laboratory of Surface Techniques/Technologies, Institute of Metallurgy,
Faculty of Materials Science and Engineering, University of Miskolc, Miskolc, Hungary
e-mail: fektt@uni-miskolc.hu
G. Lassú
e-mail: feklassu@uni-miskolc.hu

© Springer International Publishing AG 2017 135

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_14
136 T.I. Török and G. Lassú

Fig. 1 The two major types of surface treatments, i.e. so-called surface engineering techniques
widely used to finish many automotive parts [1]

highly sophisticated destructive type techniques. One such and very important
property is the elementary composition of the surface zones which can be analysed
in depth with high accuracy and in a rather fast manner by a special surface
analytical spectroscopic technique called Glow Discharge Optical Emission
Spectroscopy (GD OES).

2 Glow Discharge Optical Emission Spectrometry

(GD-OES)

The phenomenon of cathode sputtering is known for a long time, but the use of
glow discharge (GD) in spectrometry started when Grimm [2] built his first glow
discharge spectrometer with a hollow anode source. By now the glow discharge
spectrometry has become a widely used tool for surface and interface analysis with
the Grimm type glow discharge source. The Grimm glow discharge source is a flat
type source, which consists of an anode tube and a flat sample playing the role of
the cathode. There is a spacer (ceramic cathode block and O-ring) between the
sample surface and the anode tube to maintain a fixed distance (d = 0.1–0.3 mm)
and assure the vacuum tightness [3]. The volume in front of the flat sample is
Utilization of the GD OES Depth Profiling … 137

Fig. 2 Schematic of the Grimm-type glow discharge source attached to the sample sealed with an
O-ring. A circular spot with a diameter of 4 mm is sputtered with high purity argon forming a
crater the bottom of which is so analysed continuously in depth

pumped down to the vacuum *0.1–1 Pa then filled up by high purity Ar gas up to
300–1300 Pa. (See Fig. 2.)
The flat sample properly attached to the Grimm-type GD source is then sputtered
and atomized by the high purity argon plasma and the constituting atoms of the
sample so ablated into the plasma will emit light which is detected by the optical
emission spectrometer (OES). In this way and using RF (radio frequency) plasma
excitation one can sputter and erode electrically conductive and non-conductive
materials alike, which is a great advantage of the given system. For demonstrating
this special attribute of our RF powered GD OES apparatus, the first example is a
spectrum recorded while analysing a vitreous enamelled steel sheet (Figs. 3 and 4).
In the qualitative GD OES spectrum (Fig. 3) all the major constituting elements
(Na, Si, O; Ca, Ba, K, Al, B etc.) of this ceramic (silicate) type coating could be
detected and analysed down to the base steel sheet (metal substrate) where the
signal of iron (Fe) is started increasing at a crated depth of around 140 µm, and it
took only about 25 min sputtering, i.e. testing time.
The quantitative GD in depth spectrum (Fig. 4) reveals even more information
about the distribution of the elements of the coating, for example, the increasing
concentrations of iron closer to the substrate (between about 100–140 µm) is the
consequence of the evolution of a bonding zone between the steel and the
enamel-coating.
Otherwise, it is worth mentioning here that using such ceramic coatings on some
automotive parts, like on cast exhaust manifolds is becoming a novel and
138 T.I. Török and G. Lassú

Fig. 3 GD OES depth profile analysis of a glassy (vitreous) enamel coated steel sample. Light
intensities of the elementary constituents (detected by the photomultipliers and measured in volts)
are shown in the function of both the sputtering time and crater depth as the sample was
ablated/eroded by the argon plasma

Fig. 4 Quantitative in depth elementary composition (in weight percentages) of the same vitreous
enamel coated sample as in Fig. 3
Utilization of the GD OES Depth Profiling … 139

acceptable idea of today [4]. And so, the traditional vitreous (glassy) enamels are
finding new applications in the modern car making industry as a special and highly
heat and corrosion resistant composite coating material.

3 Testing Multilayered Coatings (Car Body Panels)

with GD-OES Depth Profiling

Car body panels of today can be viewed as a well-known example of the very
successful application of one of the most advanced surface coating technologies
applied both for protection and aesthetical reasons. Such multilayered coating
systems consist of several thin sublayers as it is shown in Fig. 5.
As the overall thickness of such car body coatings can be much above 0.1 mm,
which is close to the maximum crater depth accessible by our GD OES apparatus,
so some of the upper part of the coating must be previously removed prior to
starting the GD sputtering if one wishes to analyse the multilayers down and/or into
the substrate as well of such samples [6]. The present GD OES record (See in
Fig. 6) was obtained by analysing a painted and lacquered (clearcoat) car body
panel where the in depth qualitative spectrum shows many important features of the
multilayered coating. In Fig. 6 it is clearly seen that below the organic (C, O)
clearcoat (Lacquer) there is an effect basecoat (Al), below which the detected
elements (Ba, S, Ti, O etc.) are related to the applied colouring pigments and fillers,
while the Zn and Ni peeks indicate that the steel panel must have been coated

Fig. 5 Structure of three-layer automotive coating/left/and four-layer coating with effect basecoat
and solid colour basecoat/right/[5]
140 T.I. Török and G. Lassú

Fig. 6 Qualitative in depth GD OES elementary analysis of a multilayer painted car body panel
with marking the major constituting elements down to the steel sheet (base metal)

(electroplated) with a thin zinc-nickel alloy layer as well to enhance the overall
corrosion resistance of the finished steel car body panel. Otherwise, such a surface
engineering approach is one the most sophisticated ones of today.

4 Plasma Cleaning and GD OES Testing

of the Cleanliness

As it is well-known, there are many proven and widely used washing-cleaning

methods applied in the car manufacturing industry to clear away even thick con-
taminant layers and corrosion products from automotive parts. However, more
recently there is a growing demand for the removal of even the last traces of
residual contaminants, that is, there is growing demand for ultrafine cleaning. And,
to answer that demand some newly emerging surface cleaning technologies like
plasma treatments working with partially ionized (plasma) gases are also coming to
be introduced in the car industry. Then, in connection to that, the surface cleanliness
must also be monitored with high sensitivity, for which task the GD OES surface
analytical technique can also be a suitable tool to check and evaluate the pre- and
post-treatment surface cleanliness and surface activation state [7].
Utilization of the GD OES Depth Profiling … 141

Fig. 7 The open air atmospheric plasma jet laboratory apparatus (Plasmatreater AS 400) installed
in the Surface Techniques Laboratory (Institute of Metallurgy)/left/. An industrial application
example of using plasma jet for surface fine cleaning and surface modification/activation before
gluing and sealing a car headlight housing/right/[8]

Fig. 8 Results of open air plasma jet fine cleaning monitored/tested by GD OES analysis through
checking the surface carbon (C) indicative of the surface residual contamination (or cleanliness)
level after scanning the surface with the cleaning plasma jet [9]

The effectiveness of our laboratory plasma system (shown on the left in Fig. 7)
for fine surface cleaning was demonstrated through performing repeated plasma
cleaning experiments of artificially contaminated surfaces (stained with liquid
DMSO) in several laboratory tests [8], and a kind of a visual summary of the results
is shown in Fig. 8.
Though the results presented in Fig. 8 are only qualitative, but the effectiveness
of the plasma fine cleaning is quite convincing if one compares the low levels of the
C intensities detected after scanning the original (already quite clean) surface and
that of the artificially DMSO (dimethyl sulfoxide) contaminated then plasma
cleaned one.
142 T.I. Török and G. Lassú

5 Summary

Almost all automotive parts are surface treated by one way or another, and this paper
highlights a few examples of the utilization of a highly sophisticated RF GD OES
surface analysis technique which can be used for in depth analysis of inorganic or
organic surface coatings alike as well as to check/detect the surface contamination
after fine cleaning or follow the elementary composition of the close-to-surface
zones down to about 200 µm. As the given GD OES equipment has the capacity of
simultaneously detecting more than 40 chemical elements in a very wide range
(often between 100% and 1 ppm), it can be efficiently used to analyse rather thick
nonmetallic vitreous glassy enamel layers, multilayers like that of the modern sur-
face coatings of car body panels as well as very thin “nanolayers” of residual surface
contaminations after even the most thorough traditional surface cleaning procedures.
In connection to the latter example, one novel application for fine surface cleaning
by our open air laboratory atmospheric plasma jet was also demonstrated.

Acknowledgements This research was (partially) carried out in the framework of the Center of
Applied Materials Science and Nano-Technology at the University of Miskolc. The described work
was carried out as part of the TÁMOP-4.2.2.A-11/1/KONV2012-0019 project in the framework of
the New Széchenyi Plan. The realization of this project is supported by the European Union,
co-financed by the European Social Fund.The described article/presentation/study was carried out
as part of the EFOP-3.6.1-16-00011 “Younger and Renewing University – Innovative Knowledge
City – institutional development of the University of Miskolc aiming at intelligent specialisation”
project implemented in the framework of the Szechenyi 2020 program. The realization of this
project is supported by the European Union, co-financed by the European Social Fund.

References

1. Kennedy DM (2008) Exploitation of surface engineering technology research. In: Conference

papers, Dublin Institute of Technology
2. Grimm (1967) Glimmentladungsröhre. German patent DE1589389
3. Nelis T, Pallosi J (2006) Glow discharge as a tool for surface and interface analysis. Appl
Spectroscopy Rev 41:227–258
4. Method of making a ceramic coated exhaust manifold and method, US patent 5718046 (1998)
5. Poth U (2008) Automotive coating formulation. Vincentz, Hannover (Germany), p 43
6. Török TI, Lévai G, Szabó M, Pallósi J (2010) Characterizing coatings of car body sheets by
glow discharge optical emission spectrometry (GD-OES), Chap 11, pp 335–349. In:
Makhlouf N (ed) High performance coatings for automotive and aerospace industries. N.Y.
7. Török T, Urbán P, Lassú G (2015) Surface cleaning and corrosion protection using plasma
technology. Int J Corros Scale Inhib 4(2):114–122
8. Bonding of headlight housing with Openair Plasma. https://www.youtube.com/watch?v=
V3wagCyiatg/made public on 16 Oct. 2013/
9. Urbán P, Kun É, Ferenczi T, Sós D, Török T (2014) Fine cleaning of metal surfaces by air
plasma jet treatment, BKL Kohászat. J Metall 147/2:25–28 (in Hungarian)
Analysis of Surface Topography
of Diamond Burnished Aluminium Alloy
Components

Gyula Varga and Viktória Ferencsik

Abstract Nowadays diamond burnishing, which belongs to the cold plastic man-
ufacturing procedures, is used more frequently for final finishing operations of parts.
By its application, the surface roughness and the micro-hardness in the sub-layers of
the components can be increased. The procedure of diamond burnishing can be
performed for final finishing manufacturing of outer and inner cylindrical surfaces,
and shaped surfaces (e.g. conical, spherical and even statue like) too. The parameters
which effect to the surface features during manufacturing are burnishing speed, feed
rate, burnishing force, the number of passes, material and geometrical data of the
working part of the burnishing tool, furthermore the lubricant applied to burnishing.
During our experiments we have chosen from the above mentioned parameters the
burnishing speed, the feed rate and the burnishing force and we examined what is the
effect of these parameters to the surface topography when manufacturing outer
surface of cylindrical components by burnishing tool having given geometrical
dimensions. The experiments were executed by the Factorial Experiment Design
method. On the base of the evaluated experiment data the improvement ratio of
surface roughness was determined by empirical formulas. The technological
parameter and burnishing force values were shown out, which provided the highest
improvement ratio of surface roughness.

1 Introduction

Cold plastic manufacturing is a dynamically developing forming technology. The

main endeavour of modern plastic forming is to create the shape and dimension of
the planned part accurate, next to the assuring of appropriate values of strength and
strain characteristic. This process does not require high volume of coolants and

G. Varga (&)
V. Ferencsik

University of Miskolc, Miskolc, Hungary
e-mail: gyula.varga@uni-miskolc.hu
V. Ferencsik
e-mail: ferencsik.viktoria@uni-miskolc.hu

© Springer International Publishing AG 2017 143

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_15
144 G. Varga and V. Ferencsik

lubricants, so that type of manufacturing operation can be realised which envi-

ronmental load is very small. The surface strain hardening belongs to these pro-
cedures, which is characterised by cold hardening and residual stresses [1].
The technological solutions of mechanical surface toughening are differing from
each other in the feature of relative displacement of the working surfaces of the
workpiece and the tool. From this point of view, it can be differed:
• burnishing, applying sliding relative displacement,
• roller burnishing, based on the rolling relative displacement, and
• surface hardening technological procedures applying impact body combined
with dynamic effects.
The main types of the above mentioned procedures are shown in Fig. 1.

(a) (b)

(c)

(d)

Fig. 1 Tools for cold working operations: a single-roller mechanical tools, b multi-roller
mechanical tool, c deep rolling tool, d burnishing tools [2]
Analysis of Surface Topography of Diamond Burnished … 145

Life time of workpieces can be reached economically by different precision

finish machining procedures e.g. precision grinding [3]. Among the manufacturing
procedures we intend to examine the diamond burnishing. If diamond burnishing is
used in manufacturing companies, it requires the environmental oriented monitor-
ing [4]. By the application of wear-resistant complexes for modifying its surface
properties the improvement of working efficiency of cutting tools can be deter-
mined [5]. In the innovative manufacturing processes, the evaluation of machine
tool quality is a very important activity [6] and different machining procedures such
as hard turning [7], next to it the optimization of milling manufacturing operation
[8, 9].
Several studies deal with the theoretical and experimental examination of bur-
nishing process, more exactly, how the different burnishing parameters have effects
on surface roughness of burnished part.
During their examinations Luo et al. dealt with diamond burnishing of com-
ponents made from aluminium alloy type LY12 and bronze type H62 [10, 11].
They created empirical formulas and experienced that from the point of view
micro-hardness of the burnished surface the depth of burnishing is the most
effective parameter which is followed by the burnishing force, feed rate and the
speed of burnishing. Among the burnishing force, feed rate and burnishing speed
Yu and Wang took into account the effect of nose radius of the tool, furthermore the
rigidity of the tool spring to the surface roughness when executed experiments on
aluminium alloy [12]. According to the obtained results, it was stated that the use
of greater tool noise radius and higher tool spring rigidity is more advantageous.
Stalin John et al. applied the low plasticity burnishing (LPB) [13] to their experi-
ments those workpiece materials which widely applied in the industry as EN24,
AISI 1020, Al606, and bronze.
According to the examinations related to surface micro-hardness and circularity
generally it can be stated that the increase of the feed has an unfavourable effect on
both of them. Tadic et al. experimentally examined the rigidity of burnishing tool
when the material of the burnished workpice was aluminium alloy, type
EN-AW-6082 [14]. Their experiments verified their teoretic statement: among the
surface roughness and burnishing parameters, as burnishing force, feed rate and
number of passes, the extents of the correlations are high. Majzoobi, Jouneghani
and Khademi examined the lifetime [15] against fatigue by numeric simulation of
material type Al7075. They concluded that by increasing of the number of passes
and decreasing the feed rate at the same time the depth of zone of residual stress and
its homogeneity is increasing. Randjelovic et al. approached the examination of
burnishing procedure theoretically by applying numeric simulation [16]. For cre-
ating of Finite Element Model they took the kinematics of peripheral milling as a
basis and their examined material was aluminium alloy type EC1350. They stated
that the best surface roughness can be achieved when the burnishing interpene-
tration depth almost approaches the maximum peak height (Rp).
Taking into account the literatures mentioned above, the Taguchi type full
experiment design method [17] was used in our investigations when burnishing
aluminium alloy workpiece. The aim of the examinations is to reveal connection
146 G. Varga and V. Ferencsik

among burnishing parameters and surface roughness in case of workpieces made

from aluminium alloy.

2 Experimental Investigations

2.1 Workpiece Characteristics

The material and the hardness of the workpiece to be burnished can be differed for a
very wide range. For the experiments we have chosen lightly alloyed aluminium
material. The purpose of examination of aluminium alloy material was that
non-ferrous materials such as aluminium alloys are more often applied in a variety
of industrial sectors such as automobile, aerospace and astronautics. The reason of
this, is that most of the non-ferrous materials have low density and have good
mechanical properties [10]. The examination of the chemical composition of the
lightly alloyed aluminium was executed on a scanning electron microscope type
Apollo X. The results of the measurements on three points are shown in Table 1,
where: Wt (%)—weight ratio, At (%)—atomic ratio.
The shape and dimension of the workpiece are shown in Fig. 2.

2.2 Burnishing of Outer Cylindrical Surfaces

In diamond burnishing the reduction of surface roughness and improving of

the shape correctness of the surface is caused by the interaction between the
tool, - which material is much harder than the workpiece -, and the workpiece
while performing rotational and linear movement with sliding friction between
them [18] (Fig. 3). Generally, the material of the working part of the burnishing tool
can be hardened steel, carbide, mineral ceramics, and natural or artificial diamond.
The working part of the burnishing tool made from artificial polycrystalline

Table 1 Measured values of chemical composition of the aluminium alloy

Measurement 1 Measurement 2 Measurement 3
Elements Wt (%) At (%) Elements Wt (%) At (%) Elements Wt (%) At (%)
Al 92.97 97.05 Al 92.36 96.82 Al 91.00 96.23
Si 0.25 0.25 Si 0.19 0.19 Si 0.15 0.15
Fe 0.73 0.47 Fe 0.68 0.34 Fe 1.12 0.57
Cu 5.01 2.13 Cu 5.59 2.49 Cu 6.36 2.86
Bi 0.41 0.02 Bi 0.48 0.06 Bi 0.49 0.07
Pb 0.63 0.08 Pb 0.71 0.10 Pb 0.89 0.12
Analysis of Surface Topography of Diamond Burnished … 147

Fig. 2 Geometrical dimensions of the workpiece

Fig. 3 Schematic illustration of burnishing process [5]

diamond in our experiments with the radius R = 3.5 mm. Big advantage of using
this type of tool is that it provides higher stability for the manufacturing system
[11].
The pressure necessary for realizing of cold shaping originates from the over-
lapping being in between the working part of the tool and the surface of the
workpiece to be formed.
Plastic deformation is realised in 0.01–0.2 mm thick layer in the subsurface of
the workpiece because of the static contact shaping element (burnishing tool) and
the outer surface of the workpiece [19, 20].
Burnishing of outer cylindrical surfaces can be performed on conventional
universal lathes or up-to-date CNC lathes [21, 22]. During our burnishing experi-
ments a CNC lathe with flatbed produced by the firm OPTIMUM type OPTIturn
148 G. Varga and V. Ferencsik

Fig. 4 Execution of diamond

burnishing process

Table 2 Applied burnishing parameters

Sign of experiments Burnishing parameters Transformed
parameters
Fb (N) f (mm/rev) vb (m/min) x1 x2 x3
1 10 0.001 15 −1 −1 −1
2 20 0.001 15 +1 −1 −1
3 10 0.005 15 −1 +1 −1
4 20 0.005 15 +1 +1 −1
5 10 0.001 30 −1 −1 +1
6 20 0.001 30 +1 −1 +1
7 10 0.005 30 −1 +1 +1
8 20 0.005 30 +1 +1 +1

S600 (Fig. 4) was used. During our burnishing experiments the applied oil kine-
matic viscosity was m = 70 mm2/s.

2.3 The Applied Burnishing Parameters

Examined parameters: burnishing force, feed rate, and burnishing speed, which
parameter ranges are as follows:

Burnishing force Fb1 = 10 N and Fb2 = 20 N

Feed rate f1 = 0.001 mm/rev and f2 = 0.005 mm/rev
Burnishing speed vb1 = 15 m/min and vb2 = 30 m/min

The matrix of the Taguchi type Factorial Experiment Design can be seen in
Table 2, which contains the burnishing parameters in natural dimensions and their
transformed values as well.
Analysis of Surface Topography of Diamond Burnished … 149

During the examination the improvement of the surface roughness parameter

(qRa) of the Arithmetical mean deviation of the profile (Ra) was created which can
be seen in the formula (1):

Rab
qRa ¼
100 %; ð1Þ
Raa

where:
qRa the improvement ratio of surface roughness parameter Ra. This is a
dimensionless parameter which characterises the improvement of parameter
change because of manufacturing,
Rab Ra parameter before burnishing,
Raa Ra parameter after burnishing.
If the value of qRa is greater than 100, then the value of the Arithmetical mean
deviation of the profile (Ra) is improving in consequence of burnishing.

2.4 Measurement of Surface Roughness

Before and after burnishing the measurement of surface roughness of the specimen
were executed on a 3D surface roughness measuring machine type AltiSurf 520
(Fig. 5).

Fig. 5 Measuring of surface roughness on the AltiSurf 520 3D measuring machine

150 G. Varga and V. Ferencsik

Fig. 6 2D Surface roughness profiles of the Arithmetical mean deviation (Ra). a Experiment
number 1 (Table 2) before burnishing and b after burnishing

During measuring the optical sensor was used and the evaluation of the mea-
suring data was done by the own software of the measuring machine (PheNix). The
measurements were done in three different positions rotated by 120°. Figure 6
shows the changing of the Arithmetical mean deviation of the profile (Ra) for the
1st measuring set up (Table 2) (a) and (b). Figure 6 shows how the Arithmetical
mean deviation of the profile reduced after burnishing.

3 Results

The values of the measurements can be seen in Table 3. We emphasise again that
improvement in the burnishing process was realised when the value of qRa is greater
than 100.
From the values of Table 3 those burnishing parameters could be determined
where in most cases the value of the surface roughness of the burnished specimen
improved related to its state before burnishing. By the use of Factorial Experiment
Design empirical formulas can be determined [21]. Calculations and demonstra-
tions of the functions were done by the use of MathCad software.
Analysis of Surface Topography of Diamond Burnished … 151

Table 3 Measured values of the surface roughness Ra

Sign of the workpiece Ra (lm) qRa
After turning After burnishing
1/1 1.0151 1.0117 0.5339 0.4231 239.11
0.9324 0.4337
1.0877 0.3016
1/2 0.8596 0.9299 0.2081 0.2631 353.44
0.9703 0.3432
0.9599 0.2380
1/3 0.9117 0.9374 0.2688 0.3040 308.36
1.0416 0.3299
0.8589 0.3134
1/4 0.8977 0.8834 0.2610 0.3017 292.81
0.8745 0.2966
0.8780 0.3476
1/5 0.9472 0.9524 0.3677 0.4891 194.73
0.9423 0.3533
0.9676 0.7462
2/1 1.0228 1.1319 0.3010 0.4141 273.34
1.1294 0.4605
1.2434 0.4807
2/2 1.1285 1.0559 0.7023 0.6421 164.44
1.0405 0.5868
0.9986 0.6371
2/3 0.9532 0.9814 1.3605 1.2703 77.26
1.0050 1.1822
0.9860 1.2681

qRa ¼ k0Ra þ k1Ra Fb þ k2Ra f þ k3Ra vb þ k12

Ra
Fb
f þ k23
Ra
Fb
vb þ k23
Ra
f
vb
ð2Þ
þ k123
Ra
F b
f
vb

where:

k0Ra ¼ 67:75 k1Ra ¼ 17:354 k2Ra ¼ 6:569
104

k3Ra ¼ 0:483 Ra
k12 ¼ 2:349
3
10 Ra
k13 ¼ 0:178
Ra
k23 ¼ 1:061
103 Ra
k123 ¼ 59:85

Figure 7 shows the extent of the improvement of the application of sliding

burnishing. It can be stated that the application of the smaller burnishing speed
(vb = 15 m/min) served the higher improvements. When the burnishing force was
Fb2 = 20 N and the burnishing speed vb1 = 15 m/min than the reduction of feed
152 G. Varga and V. Ferencsik

Fig. 7 Evaluated experimental result

rate from the value of f2 = 0.005 mm/rev to f1 = 0.001 mm/rev caused a 60.63%
increase in the value of improvement ratio surface roughness parameter (qRa).
When the value of feed rate was f1 = 0.001 mm/rev, and the burnishing speed
vb1 = 15 m/min than the increase of the value of improvement ratio surface
roughness parameter (qRa) was 114.33% while the burnishing force was increased
from the value of Fb1 = 10 N to Fb2 = 20 N.

4 Summary

The paper deals with the experimental analysis of sliding burnishing when the
material of the workpiece was lightly alloyed aluminium material. Experimental
parameters were the burnishing force, feed rate and the burnishing speed.
The aim of the experiments was how these parameters have effect on the
Arithmetical mean deviation of the profile (Ra). The experiments were executed
and evaluated on the base of Taguchi type full Factorial Experiment Design. The
evaluation was more visible by the use of the dimensionless improvement ratio of
surface roughness parameter (qRa). After determining the empirical formula its 3D
demonstration made the evaluation more spectacular.
On the base of the present research work, it can be stated:
Analysis of Surface Topography of Diamond Burnished … 153

• Among the examined parameters, the effect of the burnishing speed is the most
dominant, as while the burnishing force was Fb2 = 20 N and the burnishing feed
rate f2 = 0. 005 mm/rev than the reduction of burnishing speed from the value
of vb1 = 30 m/min to vb1 = 15 m/min caused a 215.58% increase in the value of
improvement ratio surface roughness parameter (qRa).
• The most appropriate improvement ratio surface roughness parameter (qRa)
resulted when the burnishing parameters were as follows:
– Fb2 = 20 N,
– f1 = 0.001 mm/rev,
– vb1 = 15 m/min.

Acknowledgements The described study was carried out as part of the EFOP-3.6.1-16-00011
“Younger and Renewing University—Innovative Knowledge City—institutional development of
the University of Miskolc aiming at intelligent specialisation” project implemented in the
framework of the Szechenyi 2020 program. The realization of this project is supported by the
European Union, co-financed by the European Social Fund.

References

1. Tisza M, Balogh A, Schäffer J (2007) Mechanical technologies. Miskolc (Hungary), p 251

2. http://www.transet-tool.com/index.php?option=com_content&task=view&id=43&Itemid=38.
Accessed 21 June 2016
3. Kundrák J, Markopoulos A (2016) FEM/AI models for the simulation of precision grinding.
Manuf Technol 16(2):384–390
4. Rusko M, Kralikova R (2013) Implementation of environmental oriented monitoring in the
manufacturing company. Adv Mater Res 816–817:1225–1230
5. Vereschaka AA (2013) Improvement of working efficiency of cutting tools by modifying its
surface properties by application of wear-resistant complexes. Adv Mater Res 712–715:347–351
6. Kuric I (2011) Evaluation of machine tool quality. Int J Quality Res 5(4):269–275
7. Zębala W, Kowalczyk R (2014) Cutting data influence on cutting forces and surface finish
during sintered carbide turning. Key Eng Mater 581:148–153. ISSN 1013-9826. doi:10.4028/
www.scientific.net/KEM.581.148
8. Lancea C, Ivan NV, Chicoş LA, Oancea Gh (2008) Optimization of CNC milling files since
CAD phases. In: Annals of DAAAM for 2008 & Proceedings 19th international DAAAM
symposium intelligent manufacturing and automation: focus on next generation of intelligent
systems and solutions, Oct. 22–25, Trnava, Slovakia, ISSN:1726-9679, pp 741–742
9. Kundrák J, Felhő Cs (2016) 3D roughness parameters of surfaces face milled by special tools.
Manuf Technol 16(3):532–538
10. Luo H, Liu J, Wang L, Zhong Q (2006) The effect of burnishing parameters on burnishing
force and surface micro-hardness. Int J Adv Manuf Technol 28(7):707–713
11. Luo H, Liu J, Wang L, Zhong Q (2005) Investigation of the burnishing process with PCD tool
on non-ferrous metals. Int J Adv Manuf Technol 25:454–459
12. Yu L, Wang L (1999) Effect of various parameters on the surface roughness of an aluminium
alloy burnished with a spherical surfaced polycrystalline diamond tool. Int J Mach Tools
Manuf 39:459–469
13. Stalin John MR, Suresh P, Raguraman D, Vinayagam BK (2014) Surface characteristics of
low plasticity burnishing for different materials using lathe. Arab J Sci Eng 39:3209–3216
154 G. Varga and V. Ferencsik

14. Tadic B, Todorovic MP, Luzanin O, Miljanovic D, Jeremic MB, Bogdanovic B, Vukelic D
(2013) Using specially designed high-stiffness burnishing tool to achieve high-quality surface
finish. Int J Adv Manuf Technol 67:601–611
15. Majzoobi GH, Zare Jouneghani F, Khademi E (2016) Experimental and numerical studies on
the effect of deep rolling on bending fretting fatigue resistance of Al7075. Int J Adv Manuf
Technol 8(9):2137–2148. doi:10.1007/s00170-015-7542-z
16. Randjelovic S, Tadic B, Todorovic MP, Vukelic D, Milarodovic D, Radenkovic M, Tsiafis C
(2015) Modelling of the ball burnishing process with a high-stiffness tool. Int J Adv Manuf
Technol 81(9):1509–1518. doi:10.1007/s00170-015-7319-4
17. Fridrik L (1989) Chosen chapters from the topic of planning of experiments of production
engineering. Budapest (Hungary), pp 109
18. Akkurt A (2011) Comparison of roller burnishing and other methods of finishing treatment of
the surface of openings in parts from tool steel D3 for cold forming, Metal Science and Heat
Treatment 53(3–4), (Russian Orig. Nos. 3–4, March–April, 2011), pp 145–150
19. Luca L, Neagu-Ventzel S, Marinescu I (2005) Effects of working parameters on surface finish
in ball-burnishing of hardened steels. Precision Eng (Elsevier) 29:253–256. doi:10.1016/j.
precisioneng.2004.02.002
20. El-Taweel TA, El-Axir MH (2009) Analysis and optimization of the ball burnishing process
through the Taguchi technique. Int J Adv Manuf Technol 41:301–310. doi:10.1007/s00170-
008-1485-6
21. Varga G (2016) Possibility to increase the life time of surfaces on parts by the use of diamond
burnishing process. Key Eng Mater 686:100–107. ISSN:1662-9795. doi: 10.4028/www.
scientific.net/KEM.686.100
22. Varga G, Sovilj B, Pásztor I (2013) Experimental analysis of sliding burnishing. Acad J
Manuf Eng Editura Politechnica 11(3):6–11. ISSN:1583-7904
Investigation of Tyre Recycling
Possibilities with Cracking Process

Viktória Mikáczó, Andor Zsemberi, Zoltán Siménfalvi

and Árpád Bence Palotás

Abstract The field of vehicle tyres is a key pillar to the Vehicle Engineering BSc
launched in September 2016 at the Faculty of Mechanical Engineering and
Informatics of the University of Miskolc and the Tyre manufacturing postgraduate
course in the technological specialisation on which work is in progress. Generating
a yearly amount of several 100 millions of tyres as waste of the automotive industry
is, almost 80% of them as passenger car tyres and 20% as truck tyres, whose
management creates a huge load to bear on society. These days a relevant task of
this field is to find a solution that is reducing environmental loads and sustainable to
the solution. Vehicle tyres contain many organic and inorganic compounds: natural
and artificial caoutchoucs (NR, SBR, BR, IIR, EPDM), silica, zinc oxide, sulphur,
steel and artificial fibres, anti-ageing agents, carbon black etc. whose production
requires a significant use of fossil energy carriers. There are several ways of
recycling tyres lost their original function: incineration, recycling in its material
(rubber-based pavements, roads, sporting grounds) or chemical conversion (energy
carrier, chemical raw material), respectively. These days cracking in combined
material flow embodies one of the main research directions of chemical conversion.
The bottom line is that several raw materials are decomposed in parallel during
catalyst-assisted thermal cracking: blends of different ratios of biomass, plastics,
rubber tyre. This publication presents options of chemical conversion and its

V. Mikáczó (&)
A. Zsemberi
Z. Siménfalvi

Institute of Energy Engineering and Chemical Machinery,
University of Miskolc, Miskolc, Hungary
e-mail: mikaczo.viktoria@uni-miskolc.hu
A. Zsemberi
e-mail: zsemberi.a@gmail.com
Z. Siménfalvi
e-mail: simenfalvi@uni-miskolc.hu
Á.B. Palotás
Department of Combustion Technology, Institute of Energy and Quality
Affairs, University of Miskolc, Miskolc, Hungary
e-mail: arpad.palotas@uni-miskolc.hu

© Springer International Publishing AG 2017 155

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_16
156 V. Mikáczó et al.

optimisable parameters. We investigated thermo-catalytic thermal cracking

(cracking) of rubber and polystyrene waste producing thereby valuable petrol- and
gas oil-type hydrocarbon products.

1 Introduction

Entailing many enterprises and organisations, automotive industry covers the fields
of design, development, manufacturing, marketing and sales in relation to a broad
scale of vehicles used in a number of walks of life. It is one of the most lucrative
economic sectors, which has undergone a rapid evolution after the world wars and,
although set back by the oil crises of the 1970s, is still in constant progress in our
days. Hungary also has had a vivid presence in this process: various automotive
associations have been established in the last years (such as Bakony-Balaton
Mechatronics and Automotive Cluster) and large car makers consider Hungary as a
potential labour market target that played a huge role in the improvement of the
domestic export already in 2011. In response to the need for expertise arising in
relation to these, the Faculty of Mechanical Engineering and Informatics of the
University of Miskolc elaborated and launched the curriculum of the Vehicle
Engineering BSc starting in September 2016, moreover, the Tyre manufacturing
postgraduate course in the technological specialisation on which work is in pro-
gress, offers professionals coming from other technical areas advantageous orien-
tation opportunities. The automotive industry as a leading industry branch and
along with this, the field of vehicle tyres is a relevant pillar of the two previous
courses.
The use of various artificial polymer derivatives has risen significantly in the last
100 years, the issue is, however, that several types of hydrocarbon derivatives are
increasingly prevalent in the resulted waste. It is estimated that 1.6–1.8 billion
rubber tyres and 350–370 million tonnes of plastic waste were generated globally in
2015. Their raw material being crude oil, due to also this it is important to consider
polymers that have become waste to be a secondary source of raw materials as
thereby the environmental load of harmful landfills on nature can be reduced and
global CO2 emission diminished as well.
The main research direction of recycling rubber tyres is thermo-catalytic con-
version processes for energy purposes. However, it is a relevant aspect that tyres
have many materials of different quality and quantity such as styrene-butadiene
artificial caoutchouc, butyl caoutchouc, natural caoutchouc, butadiene caoutchouc,
ethylene-propylene-diene caoutchouc, zinc oxide, metal oxides, sulphur, silica,
barium sulphate, carbon black, stearic acid, IPPD, phenolic resin, CSB, hexamine,
textile (polyester, polyamide, aramid), steel.
Due to the many heterogeneous components, utilisation of rubber tyres by means
of thermo-catalytic thermal cracking is a highly complicated task posing a serious
challenge to both researchers and engineers of the current era.
Investigation of Tyre Recycling Possibilities … 157

2 Energetical Significance of Recycling

Mankind obtains the majority of its energy demand mainly from fossil energy
carriers. The percentage distribution of the world’s energy demand broken down
into sources is shown in Fig. 1.
Figure 1 reveals that only 13.8% of the produced energy arises from renewable
sources that imply an ever increasing threat to nature. Another relevant aspect is
that motorisation in developing countries (e.g. China, India) is on a hectic rise
because of which hydrocarbon-based engine fuels are demanding more and more,
covered today mainly still from crude oil.
Mankind has accumulated a rubber waste and plastic waste base of several
thousand billion tons in the last 100 years that can be converted in a so-called
cracking in combined material flow having adequately optimised operational
parameters into a high-quality liquid energy carrier.

2.1 Products of Cracking of Rubber Tyre

By means of thermo-catalytic process of rubber tyre, energy carriers and many raw
materials for the chemical industry can be produced: gas products (methane, ethane,
propane, hydrogen), petrol-, gas oil- as well as heavy oil-type hydrocarbon frac-
tions. Based on professional literature data, mostly the opportunity of utilising in
energetics emerges regarding further utilisation [1].
The economic competitiveness of these technologies can be considerably
improved by increasing the yield of so-called volatile products (gas, petrol-, gas,
oil-type fractions) therefore our publication focuses on this area as well. It can be
said on the basis of the above thoughts that the hidden exploitation of polymer
derivatives by the formed valuable products can promote the contribution to ensure

Fig. 1 Percentage
distribution of the world’s
energy demand, according to
sources (2015); US
Department of Energy
158 V. Mikáczó et al.

the increasing energy demand and thereby reducing the number of landfills extre-
mely harmful to nature.

2.1.1 Challenges of the Cracking Process

The liquid product produced purely from rubber tyre has a significant content in
olefin, which is mainly due to the fact that the macromolecules of caoutchouc are
unsaturated carbon chains. The most relevant disadvantages of liquid fraction
produced via the process:
• High content in olefin: causes thermal instability inhibiting the storage possi-
bilities of bio oil.
• High content in ash and solids.
• The produced medium is corrosive.
• High content in heteroatoms.
The latest worldwide research activities aimed also at being able to improve
disadvantageous properties already during the cracking process so that the eco-
nomic competitiveness of the technology can be raised too. Such an alternative
solution is the so-called thermo-catalytic thermal cracking in combined material
flow researched by us too during which the plastic admixtures at an adequate ratio
to rubber waste, improves the quality of generated liquid fraction due to its sig-
nificant hydrogen content.

3 Recycling Options of Rubber Tyres and Polymer

Derivatives

Environmental requirements and the related measures at many points of the world
do not work properly because of difficult and often non-transparent economic sit-
uation. Ever more stringent regulations with regard to waste management come into
force in the European Union. Member states are expected to carry out technological
developments that are able to manage the increasingly more voluminous waste to
obtain the best efficiency.

3.1 Mechanical Recycling of Waste

The mechanical method can be used for recycling a material flow composed of a
homogeneous polymer. These types of processes require high-purity collection or
precise selection therefore their application is not overly popular. After proper
pre-treatment and selection, both process and real polymer waste can be recycled by
Investigation of Tyre Recycling Possibilities … 159

mechanical recycling. Raw materials undergo first shredding as the form of powder
or pellet is more advantageous in further processing.

3.2 Waste Incineration

Incineration is an exothermic process during which the organic content of the waste
is turned into gases and water by burning into water leaving the combustion
chamber as flue gas. The non-combustible materials remain as ash or fly ash. Such
procedures may contribute to the achievement of volume reduction of 85–95% of
starting waste and mass reduction of 60–70% on average. The time-frame of the
method is relatively low, but its disadvantage is that its investment and operational
costs come are high.

3.3 Chemical Utilization of Polymers

During chemical utilisation polymers are converted into shorter molecules thus, in
general liquid and gas products develop. The product fractions can be used both as
a petrochemical raw material or a fuel. With the appropriate knowledge of the
kinetics of these processes has been often incomplete, thus, the production of the
product of the desired quality poses a high challenge as we already mentioned
before.
Caoutchoucs of various types and polyethylene, polypropylene and polystyrene
have several advantageous properties arising from their properties for chemical
recycling. However, it presents difficulties if the raw material to be processed has
contaminants as these can appear in the products as well. Such can be physical
contaminations (dust, oil, etc.) occurring on the surface of rubber and plastic waste
or hetero atoms in the structure of polymer (O, S, Cl etc.).

3.3.1 Theory of Thermal and Thermo-Catalytic Thermal Cracking

Cracking and pyrolysis can be a purely thermal or thermo-catalytic process, in

which the long carbon chains (polymers) break up into a shorter carbon chain
saturated and unsaturated hydrocarbons by cracking of C, C- and C, H-bondings. It
is important that thermal cracking performed below 650 °C is called cracking,
whereas processes above 650 °C are called pyrolysis [2].
Catalytic cracking and pyrolysis technologies employ principally
aluminium-silicate (zeolite) type catalysts, thus, it is accompanied by isomerisation,
ring closure and hydrogen transfer as well. The products obtained so contain mainly
more valuable branched alkanes (isoparaffins) as well as hydrocarbons and only
some small quantities of alkenes occur [3]. In the thermal cracking processes
160 V. Mikáczó et al.

performed below 650 °C mostly a complicated mixture of liquid hydrocarbons

develops, therefore they can be more relevant for mainly the preparation of engine
fuels [2]. In summary, the below processes must be reckoned with during the
thermal and thermo-catalytic thermal cracking [3]:
• Initiation, during which radicals, or ions come into being.
• Formation of further unstable compounds, targeted or random chain cleavage:
depolymerisation, formation of monomers, hydrogen-transfer reactions, inter-
molecular hydrogen migration (formation of paraffin and diene), vinyl
group-assisted isomerisation.
• Ring closure by the recombination, the disproportion of formed radicals and
ions.
During the catalyst-assisted thermal degradation of hydrocarbons, the processes
resulting in the deterioration of polymer structure occurring due to the thermal and
catalytic effect competing with each other [3].

3.3.2 Composition of Rubber Tyres

The application area of rubber tyres and various polymer derivatives range across a
broad spectrum. Therefore, to produce an oil of adequate quality, the composition
of potential raw materials must be researched too.
Table 1 presents the material composition of rubber tyres of an average pas-
senger car and a truck [4].
Both passenger car and truck tyres are the outcome of highly complex engi-
neering activities designed to meet dynamic loads from aggressive environmental
and application on the long term and safely. Table 1 reveals as well that
carbon-based material to be used as raw material for cracking makes up 67–74% of
the whole mass. That means that maximum 50–55 kg liquid hydrocarbon can be
produced from 100 kg of rubber tyre that—also arising from the character of the
raw material can have many non-desired contaminants.

Table 1 Material composition of rubber tyres of an average passenger car and a truck [4]
Components Passenger car tyre (m/m%) Truck tyre (m/m%)
Rubber/elastomer 47 45
Carbon black 21.5 22
Metal 16.5 25
Textile (synthetic fibre) 5.5 –
Zinc oxide 1 2
Sulphur 1 1
Fillers 7.5 5
Carbon-based substances, in total 74 67
Investigation of Tyre Recycling Possibilities … 161

4 Most Relevant Parameters of Thermal Cracking

Processes

Thermal cracking (cracking, pyrolysis) is carried out in a suitably designed reactor

in oxygen-poor or oxygen-free atmosphere, during which the material decomposes
chemically. The simplified process of thermal cracking is illustrated in Fig. 2.

Fig. 2 The simplified workflow of thermal cracking process

162 V. Mikáczó et al.

During the processes executing thermal cracking the following operational

variables play the most important role [5, 6]:
• Temperature
• Residence time
• Operating pressure
• Selection of catalyst
• Mixing conditions
• Granular size
• Heating rate
Their appropriate selection is indispensably important to ensure proper product
distribution. The pyrolysis is an endothermic process, therefore it is essential to
ensure proper temperature, which is one of the most important operational variables.
It can be generally said that more gas state product develops in high temperature
processes (over 600 °C). In low temperature processes (below 400 °C) less gas
develops, but the formed fluid product is much more viscous and more secondary
product develops. The majority of the processes is made occur at 400–500 °C [7].

4.1 Effect of Temperature on Thermal Cracking Processes

Plastics, artificial caoutchoucs and other organic polymers break down into smaller
carbon chain entities as temperature rises. Simply put, this constitutes the basis for
the development of valuable volatile fractions. Without applying a catalyst, the
degradation of polymers during thermal cracking occurs via a radical mechanism,
therefore, the share of unbranded hydrocarbons shall be larger in the volatile
products compared to thermo-catalytic cases, thus the latter is more beneficial
economically as well. ‘Today and on the medium term (about until 2040) mainly
engine fuels and lubricants with high isoparaffin content shall constitute economic,
environment-friendly and safe fuels [8].
The rise in temperature has a significant effect on the decomposition of polymers
as the reaction-rate constant rises with temperature significantly. It is to be observed
that a lot more volatile product develops at a higher temperature in a much shorter
time [7].

4.2 Effect of the Catalyst in the Thermal Cracking Process

During catalytic processes decomposition processes happen via ionic mechanisms.

In the presence of a catalyst, the share of gas yield grows considerably, further, the
average molecular weight of fluid phase developed so is lower than in the thermal
Investigation of Tyre Recycling Possibilities … 163

case. This phenomenon is due to the catalyst to increase secondary cracking. The
products developed so lead to more valuable hydrocarbon fractions. It is also
beneficial that less energy must be input to the system in the catalytic process as this
decreases activation energy [7].

5 Presentation of the Experimental Process

The goal of our research was to develop an experimental equipment that is able to
analyse thermo-catalytic thermal cracking processes with an optimum parameter
combination.
By designing a so-called multi-stage catalyst attachment an opportunity opens
up to make qualitative and quantitative analysis of developed products available in
the widest possible range. The formed share of quantities of various products allows
for the conclusion of kinetic rate parameters too. Our goal in the establishment of
the adequate mathematical model is to model these processes—in a way coherent
with measurement points—using computer modelling in the future.

5.1 Decisive Processes and Components of the Experiments

Figure 3 shows the workflow of the measurement equipment. The unit marked with
number 1 is a nitrogen cylinder by means of which the system is inertialised. By
infeeding nitrogen, a slight overpressure, oxygen exclusion during the operation is
ensured as well as the residence time of product steams in the tube reactor system is
controlled.

Fig. 3 Workflow of the experimental equipment: 1. nitrogen bottle; 2. furnace_01; 3. vertical

reactor; 4. horizontal reactor (catalyst attachment); 5. furnace_02; 6. condenser; 7. liquid collector;
8. rotameter [9]
164 V. Mikáczó et al.

Fig. 4 The insulated reactor

system in top view

The raw material was infed in unit 3, the vertically positioned tube reactor. The
reactor tube was heated with Hőker Cső 250/900 (2) electrically heated furnace
whose power is 650 W. Unit 4 is a so-called horizontally positioned tube reactor
(thermo-catalytic tar decomposer) heated by operational unit 5 (Hőker Cső
250/900), which is just as an electrically heated furnace with a power of 400 W.
The selection of proper structural material is extremely important for appliances
operating at high temperatures. Based on the analytical aspects, Wnr. 1.4845 (H9)
grade austenitic heat resisting steel was selected as most appropriate for the
structural materials of the externally heated reactor (3) and thermo-catalytic tar
decomposing tube reactor and the solid particle separation unit. Figure 4 shows the
uninsulated tube reactor system along with the installed furnaces.
The right hand side in Fig. 4 the nitrogen infeed valve can be seen, which is a
very important unit of the experimental process.
The device marked with number 6 in Fig. 3 is the condenser, which cooled down
the high-temperature (350–450 °C) hydrocarbon steams and this way a liquid phase
is produced that is gathered in the liquid collector tank 7. Unit 8 is the rotameter,
which measured the quantity of gas fraction.

5.2 Presentation of Raw Materials Applied

3 parallel series of experiments were performed in the measurements each as per the
following:
• homogeneous polystyrene waste, 40 g
• homogeneous rubber waste, 40 g
• PS and rubber waste blend, 20 g each
Investigation of Tyre Recycling Possibilities … 165

Fig. 5 The zeolite catalyst before the measurements (a), the nickel oxide metal and zeolite
catalyst remained after the measurements (b)

In the measurements 5 g of a zeolite and nickel oxide-type catalyst were used by

blending it in the raw material in the vertical tube reactor. Figure 5a shows the
zeolite catalyst before the measurements. Figure 5b illustrates the nickel oxide
metal and zeolite catalyst remained after the series of measurements.
Figure 5b reveals that the catalyst during the measurements undergoes a serious
carbonisation that brings about deactivation. Because of this, catalysts must be
regenerated at 800 °C in oxygen atmosphere. The exact tendency to drop in useful
efficiency is still unknown to us. Another of our future goals for research is to reveal
the rate of deactivation quantitatively as aromatisation and isomerisation reactions
are determined by the catalyst primarily and this way it has direct influence on the
composition of the more valuable liquid fraction as well.

5.3 The Steps of Measurements

40 g raw material and 5 g catalyst were infed in the vertical tube reactor in each
case, operated at 450 °C by us. The experiments ran for 50 min, out of which
heating to operating temperature made up 22 min. The horizontal operational unit
was running at 300 °C. In the first step the deoxygenation of the reactor system was
performed by means of nitrogen. The inflow of nitrogen was constantly ensured
throughout the operation. Oil formation started between 300 and 305 °C in minutes
17–18 and lasted until minutes 35–37 in the case of polystyrene.
166 V. Mikáczó et al.

5.4 Presentation of Measurements

This publication covers the quantitative evolution of hydrocarbon fractions gen-

erated during cracking as many qualitative analysis data are not yet available for all
fractions.

5.4.1 Results of Cracking of Homogeneous Polystyrene Waste

Figure 6 shows the evolution of the quantities of fractions produced from purely
polystyrene waste.
Figure 6 shows that the quantity of gas, liquid and solid fractions developed on
average one-by-one during measurements are the following: 15, 23.6, 1.4 g, which
are in average mass%: 37.6, 58.9, 3.5 m/m%. The quantity of valuable volatile
fractions arisen as 96.5% in mass% on average that can be regarded as a particularly
good outcome. The liquid product is illustrated in Fig. 7.
The average result of gas chromatography analyses is listed in Table 2 broken
down in volume% distribution.

Fig. 6 Quantity of
hydrocarbon products
produced from purely
polystyrene waste

Fig. 7 The liquid

hydrocarbon fraction
produced from purely
polystyrene waste
Investigation of Tyre Recycling Possibilities … 167

Table 2 Average gas Component Quantity (V/V%)

composition of gas samples
produced during cracking of CO2 2.4
polystyrene (V/V%) C2H4 9.1
C2H6 1.9
H2 45.15
O2 0.2
N2 1.2
CH4 30.85
CO 9.2

Fig. 8 Quantity of
hydrocarbon products
produced from purely rubber
waste

Table 2 shows that the gas composition is a high-quality combustible hydro-

carbon fraction whose more than two third is made up by methane and hydrogen.
We have reached the conclusion due to the high hydrogen content of the gas
fraction that polystyrene blended with rubber waste can be able to ‘in situ’
hydrogenate the olefin-rich fraction in the steam phase in the catalytic horizontal
tube reactor.

5.4.2 Results of Cracking of Homogeneous Rubber Waste

Figure 8 summarises the quantities of cracking-products produced from purely

rubber waste.
As can be seen in Fig. 8, the average quantities of gas, liquid as well as solid
hydrocarbon fractions produced from rubber waste were the following: 8.4, 23.6,
8 g, which are in average mass%: 21, 59, 20 m/m%. In contrast to the cracking
products of polystyrene waste, the quantity of gas fraction has dropped signifi-
cantly, by 16.6 mass% in total. The liquid product evolved nearly identically in both
cases. The weight of the solid remains (chark, tar) increased considerably in the
case of rubber waste cracking, which is a disadvantage for it. The explanation for
the phenomenon is that rubber contains a significant amount of carbon black, UV
and thermal stabilisers as well as anti-ageing agents due to which long carbon
chains can be more difficult to crack off at the same temperature. Furthermore, it can
168 V. Mikáczó et al.

Fig. 9 Quantity of
hydrocarbon products
produced from polystyrene
and rubber waste blend

be said that the system tends to shift more towards the formation of tar due to the
unsaturated carbon chains.
We tried to optimise the quantity and quality of more valuable gas and liquid
fractions to be produced from rubber waste by admixing polystyrene.

5.4.3 Results of Polystyrene and Rubber Waste Blend

Figure 9 summarises the results of cracking in combined material flow.

As can be seen in Fig. 9, the average quantities of gas, liquid as well as solid
hydrocarbon fractions produced from polystyrene and rubber waste blend were the
following: 10.1, 26, 3.9 g, which are in average mass%: 25.17, 65, 9.83 m/m%.
As can be seen from measurement results, the weight of solid decomposition
remains (chark, tar) can be reduced by the combined cracking of rubber and
polystyrene waste—in comparison to homogeneous rubber fraction while the
quantity of valuable gas and liquid products grows. The reason for the phenomenon
can be that hydrogen-richer polystyrene raw material hydrogenates unsaturated
carbon chains at high temperature in the presence of a catalyst and this way the
quantity of tar and chark drops in the reactor body.
The key goal of our future research is furthermore to put the formed petrol- and
gas oil-type hydrocarbon fractions to a qualitative analysis in each case and on a
broad scale. We have not shed light on how polystyrene admixed to rubber waste
affects quality parameters precisely and according to what mechanism.

6 Summary

Our work composed of researching recycling options of various polymer deriva-

tives (rubber and polystyrene waste) by means of thermo-catalytic degradation.
We determined that the gas fraction forming during thermo-catalytic thermal
cracking of polystyrene contains a significant amount of hydrogen that can be
Investigation of Tyre Recycling Possibilities … 169

suitable for ‘in situ’ hydrogenation of unsaturated carbon chains in the case of
rubber waste.
We shed light on that the quantity of petrol- and gas oil-type hydrocarbon
fractions produced from rubber waste could be increased if rubber waste was
converted with polystyrene in combined material flow. The quantity of gas and
liquid fraction on average rose 4.17 and 6 m/m%, respectively, in comparison to
homogeneous cases. The formation of tar and chark dropped 10.17 m/m% on
average, which can be considered a good outcome, but still too low compared to the
optimum industrial operation.

References

1. Miskolczi N, Nagy R (2012) Hydrocarbons obtained by waste plastic pyrolysis: comparative

analysis of decomposition describe by different kinetic models. In: Fuel processing technology
2. Markó L Organic Chemistry I, http://www.tankonyvtar.hu/hu/tartalom/tkt/szerves-kemia-
szerves/index.html, (Szerves Kémia I.)
3. Miskolczi N, Bartha L (2005) Investigation of thermal recycling of plastics and further
utilization of products. PhD dissertation
4. Evans A, Evans R (2006) The composition of a tyre: typical components. The Waste &
Resources Action Programme
5. Akhtar J, Amin NAS (2012) A review on operating parameters for optimum liquid oil yield in
biomass pyrolysis. Renew Sust Energy Rev 16(7):5101–5109
6. Velghe I, Carleer R, Yperman J, Schreurs S (2011) Study of the pyrolysis of municipal solid
waste for the production of valuable products. J Anal Appl Pyrol 92:366–375
7. Kaminsky W, Scheirs J (2006) Feedstock recycling and pyrolysis of waste plastics: converting
waste plastic into diesel and other fuels. Wiley series in polymer science, John Wiley & Sons,
LTD
8. Hancsók J (2008) Modern engine fuels. Public habilitation presentation, Veszprém
9. Zsemberi A, Siménfalvi KZ, Palotás AB (2016) Thermal and thermo-catalytic co-cracking.
GÉP Technical periodical of the Scientific Society of Mechanical Engineering 2016/3, vol
LXVII, pp 49–52
Utilisation of Various
Hydro-Carbon-Based Wastes
by Thermo-catalytic Conversion

Andor Zsemberi, Zoltán Siménfalvi and Árpád Bence Palotás

Abstract The global need for energy and raw materials is constantly on the rise as
mankind’s technology progresses. Due to more and more environmental load and
fossil energy carriers exhausted, processes designed for thermo-catalytic conversion
of various hydrocarbon-based wastes (plastics- and rubber waste, biomasses) and
fuels with a low calorific value (lignite, brown coal) have come into focus in the last
decades. The essence of these processes is that solid raw materials forming long
carbon chains can be converted at medium-high temperatures (410–450 °C) by
means of a special reactor system into more valuable hydrocarbon fractions of
liquid and gas state such as petrol-, gas oil-, fuel gas-type products. We examined in
our work, how low-quality rubber waste and/or brown coal, plastic waste raw
materials can be converted into better quality products—of primarily liquid state.
The problem raising a number of open points is a complicated optimisation issue as
various heterogeneous components and their content in aggressive contaminants
(sulphur, chlorine, nitrogen, oxygen, oxides, carbonates etc.) can largely affect
decomposition kinetics thus the quality and quantity of hydrocarbon products
formed so as well. This publication covers the system modelling techniques in
detail that can be used as a foundation for the basis of mathematical modelling of
high-complexity technical systems.

A. Zsemberi (&)
Z. Siménfalvi

Institute of Energy Engineering and Chemical Machinery,
University of Miskolc, Miskolc, Hungary
e-mail: zsemberi.a@gmail.com
Z. Siménfalvi
e-mail: simenfalvi@uni-miskolc.hu
Á.B. Palotás
Department of Combustion Technology, Institute of Energy and Quality Affairs,
University of Miskolc, Miskolc, Hungary
e-mail: arpad.palotas@uni-miskolc.hu

© Springer International Publishing AG 2017 171

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_17
172 A. Zsemberi et al.

1 Introduction

Fluid, solid and gas state products produced thermal-catalytically from solid
organic-based raw materials such as rubber, polymer and biomass are drawing more
and more attention globally as valuable energy carriers can be generated from them.
Their highly significant disadvantage is however, that they may contain a number of
contaminants: solid particles, toxic compounds, tar, heavy metal damps etc.
Because of this, the technical elaboration of the process poses a considerable
challenge. Many a publication back that heterogeneous components affect each
other’s thermal decomposition during operations [1].
Based on the data gathered by the Energy Agency of the USA and British
Petroleum, it can be presumed that the demand for coal, natural gas and liquid fuels
is expected to increase significantly further. The tendency is supported by Fig. 1.
Figure 1 reveals that crude oil, coal and natural gas shall may remain the main
energy sources of the world for a few decades to come, therefore it is relevant that
remaining reserves must be handled by mankind in an optimal way.
Computer and/or mathematical simulation is an indispensable element in today’s
engineering practice. These tools provide an opportunity to understand the pro-
cesses taking place better and as a result of this make manufacturing safer, more
efficient and economical. The amount of kinetic rate and mathematical publications
with regard to the topic is, however, fairly poor and only small quantities or
‘factory-grade’ raw materials are used in the investigations in most cases. Under
industrial circumstances, parameters determined this way can be employed to a
limited extent only.

10 15
Btu

Time [Year]

Fig. 1 Growing tendency of demand for coal, natural gas and liquid fuels globally [US
Department of Energy, British Petroleum]
Utilisation of Various Hydro-Carbon-Based Wastes … 173

2 Significance of Cracking Technologies

Figure 2 confirms and details the undoubted advantages of processes in combined

flow compared to direct incineration.
Besides the storability and multi-faceted applicability of energy carriers, no
dioxins or furans form in these processes [2].
As it can be seen in Fig. 2, active carbon and bio oil fractions obtained during
pyrolysis can be safely and economically stored, thus, these energy carriers can be
put to use at a later point. It merits mention that particular products can be utilised
in direct energy production in addition to their application as secondary raw
materials for the chemical industry.

2.1 Characters of the Products of Technology

The oil fraction obtained during the thermo-catalytic technology is basically a

brownish-black viscose liquid containing a number of acidic corrosive substances
as well. Due to the continuous and rapid depletion of fossil reserves, the energy
carriers produced by the combined material flow of solid rubber, plastic waste
and/or biomass fractions—mainly the volatile products—can imply a long-term,
predictable alternative energy source in the future:

Fig. 2 The thermal processes [1]

174 A. Zsemberi et al.

• Oil fraction: a mixture of petrol- and gas oil-type hydrocarbons.

• Gas products: methane, ethane, propane, ethene, propene, hydrogen etc.
• Solid products: chark, carbon black, tar.
The produced products have, however, many disadvantages, because of which
their utilisation is limited:
• High content in oxygen: causes thermal instability inhibiting the storage pos-
sibilities of bio oil.
• Low heat of combustion.
• High content in ash and solids.
• High water content (strongly acidic medium forms, mineral acids form from
heteroatoms: HCl, H2SO4 etc.).
• Corrosive: in part due to this, special structural materials needs to be applied for
processes designed for pyrolysis.
• Poor combustion properties.
The improvement of physico-chemical properties of producible products means
a challenge to researchers as this concerns a material of highly complex compo-
sition. Many a process is yet to be clarified in order to be able to produce various
hydrocarbon fractions economically and utilise safely [3]. Acidic medium is
expressively harmful as they exert a catalytic effect on the polymerisation of
unsaturated unstable compounds, thus, viscosity as well as quality oil further
deteriorates [4].

2.1.1 Most Relevant Contaminants of Products

The qualitative and quantitative detection of the most important contaminants of the
produced gas, liquid and solid fractions is relevant not only due to their usability.
By the comprehensive analysis of data it will be possible:
• To select the appropriate cleaning process (optimal, energy-efficient solution).
• To be able to select the structural material of the equipment to be applied too.
70–80 m/m% of the infed raw material in combined material flow—in the
presence of proper reactor parameters and catalyst—can be converted to oil fraction
[5], on top of which *10–10 m/m% gas and solid-state hydrocarbon fractions
develop as well. However, because of the issues covered in the previous chapter,
these cannot be directly used in internal combustion engines. Product
containing-components developing in the thermo-catalytic processes are consider-
able due to the heteroatom content of the raw material and depending on this they
contain the following to various degrees [6]:
• solid contaminations
• tar
• H 2O
Utilisation of Various Hydro-Carbon-Based Wastes … 175

• NH3
• H2S; COS; SO2
• HCl
• alkali and alkaline-earth metals, heavy metals
A number of sources confirm that the produced hydrocarbon fraction contains
the following to a certain extent (of course, depending also on the raw material):
compounds containing K, S, P, Cl, Ca, Zn, Fe, Cr, Br and Sb as well.
It is advised to remove the above mentioned components—before use—as they
damage significantly the contacting structural materials as well.

3 The Role of Applied Mathematical Models

The role of mathematical modelling is indisputable in science and modern engi-

neering applications. The computer evolution of the last 50 years allowed faster
manageability and better understanding of difficult and complicated issues for many
questions arising in practice. Chemical systems are complicated systems, changing
one element can have a significant impact on others, thus, even in the case of the
simplest existing chemical reactors one needs to know by all means the heat, mass,
and pulse-transfer processes of which almost all include chemical reactions and the
related generated/removed reaction heat too [7].

3.1 System Models

System models are a common language of different engineering fields. They are
characterised by a wide range of application areas, mutual convertibility and broad
scale. A model (models) follow(s) up technologies throughout their lifecycle.
A stationery model can be employed only during design. The control of the system,
however, means a timely activity, therefore, the relevance of system dynamics is a
key issue. The model used in final engineering practice is a formal model, which is
a mathematical one (variables, correlations). Selecting the type most suitable for the
solution to the task is not simple, it may be the result of weighing up several aspects
[7, 8].

3.1.1 A Priori Models

The most important group of models is the class of a priori model (theoretical
model, white box model etc.) whose bottom line is that they reflect the (internal)
structure of the modelled object in part or in full, all of their variables can be
assigned to a physical sense, their correlations are based on physical, chemical laws,
176 A. Zsemberi et al.

furthermore, they are able to transmit information both in space and time. They are
suitable to model both existing and imaginative objects. In design, a priori model
may be used exclusively. They can be employed in technology, design, the design
of technology control system, the design of instrumentation, process modelling,
process optimisation and the foundation of technology operations [7, 8].

3.1.2 A Posteriori Models

The other large group of system models is the class of a posteriori model (empirical
model, black box model etc.). The feature of these is that they do not reflect the
structure of the modelled object not even in part, the so-called input and output
variables have a physical content only, their correlations are formal that connect the
input and output variables, the expected adequacy can be ensured by the use of
measurement data. They can be used neither for spatial nor timely information
transmission. They are suitable for modelling a working object, therefore can be
used primarily in control. If, for one, the intervention signal changes, how the
controlled characteristic has changed, this shall explain how dynamically the sys-
tem changes [7].

3.2 Reaction Kinetics Approach of Thermal Cracking

Processes

In the case of various combined material flows it is worth examining processes

taking place between specific components and reactions taking place (at both micro
and macro level). This way the value of process products can be increased sig-
nificantly, meaning the reduction of ROI time of the scaled up processes too. To
optimise reaction parameters and understand processes taking place better many
reaction models have been created, which, however, describe mainly processes
taking place in pure homogeneous material flows. Very little information is
available on the modelling of blended and real-life waste.
The reaction kinetic description of cracking of rubber-, polymer waste and/or
coal is a fairly difficult task even separately as very complex processes happen and
significant differences emerge both in thermal and thermo-catalytic cases. The type
of reactor is a relevant aspect, but the quantity of raw material to be processed plays
an important role because of the blending ratios. Based on literature data, most of
cracking experiments are performed in batch reactors and fairly few are involved in
dealing with continuous or semi-continuous technologies. The reason for this is that
continuous yield needs raw material of higher volume and the construction of the
reactor is fairly complicated to make proper blending and heat transfer relations
happen. Furthermore, it can be said that even the removal of contaminants formed
during the process with more tendency to deposition and adhesion [3, 5].
Utilisation of Various Hydro-Carbon-Based Wastes … 177

3.2.1 Reaction Kinetic Examination of Rubber, Plastic and Biomass

Waste

Limin Zhou et al. analysed the combined cracking of various plastics (artificial
caoutchouc, HDPE, LDPE and PP) and low volatile substance-containing carbon
(LVC) by means of TGA. Experiments were carried out in a nitrogen atmosphere at
a heating rate of 20 °C/min. between temperatures of 20 and 750 °C. The degra-
dation of plastics occurs between 438 and 521 °C, whereas that of carbon between
174 and 710 °C (the reason for this is the similar molecular structure of polymer
[9]). They preassumed first-order reactions also in these cases, similarly to publi-
cations mentioned above. Just like most of researchers—they determined reaction
kinetic parameters by means of the Arrhenius equation [9]. Equation (1) means the
description of experimental results via conversion, meaning actually the mass
reduction curve.

W0  W
x¼ ; ð1Þ
W0  W1

where W0 is the initial weight of the specimen, W∞ is the final weight of the
specimen having passed cracking, W is the weight of the specimen at the specific
time. To determine reaction rate, the following Eq. (2) was employed [9]:

dx
¼ kðT Þ
f ð xÞ ¼ A
eðR
T Þ
ð1  xÞn :
E
ð2Þ
dt

The left side of the equation expresses the conversion belonging to the specific
time. This correlation was made linear such that natural logarithms of both sides
were taken, illustrated by Eq. (3) [9]:

dx E 1
lnð Þ ¼ ln A þ n
lnð1  xÞ 
: ð3Þ
dt R T

The left side of the equation was depicted versus 1/T whose slope was used to
determine the activation energy (E) and frequency factor (A) for each conversion.
First-order mechanism was presumed for the reaction, obviously [9]. Results
obtained via the mentioned methods are illustrated in Fig. 3.
The numerical value of concrete reaction kinetic parameters can be generated by
means of Fig. 3. In the light of this, the mass-, component- and heat balance of the
system can be described as well.
178 A. Zsemberi et al.

Fig. 3 Steps of determination of capitalisation energy and frequency factor [9]

4 Measured and Calculated Results

The kinetic model employed by us presumes an approach based on first-order

reaction kinetics to describe the degradation of polymers, illustrated by Eq. (4).

dW
 ¼ k
wn : ð4Þ
dt

In this relation the mass reduction curves of the polymer must be determined and
from these the general reaction rate constants can be calculated. In addition to the
latent activation energies and reaction rate constants, also the pre-exponential
factors can be determined by the quantity of the formed volatile products (gas and
liquid).
By using the adequate kinetic model, the yield and key properties of the products
can be predicted with a high degree of accuracy. The main issue is posed by the
determination of the reaction rate constant as cracking is significantly affected by
many physical, geometrical and the steric factor. The activation energy and
pre-exponential factor are to be determined by means of Eq. (5):

k ¼ A
eR
T ;
E
ð5Þ

where ‘E’ is activation energy, ‘A’ is the frequency factor (pre-exponential factor),
‘R’ is the gas constant and ‘T’ is the absolute temperature. The dimension of
frequency factor and ‘k’ reaction rate constant is the same as they are based on the
concentration unit and therefore depend on the reaction order.
Utilisation of Various Hydro-Carbon-Based Wastes … 179

Correlations may become simpler by using thermogravimetry and various

coupled methods, but in this case a new parameter needs to be introduced. This new
parameter is the heating rate.
The main problem in the cases of thermo-catalytic cracking is that the activity of
catalysts deteriorates during the process. Therefore, first-order kinetics may be used
via simplifications in such cases. According to the other option, there exists a model
for fluid catalytic cracking, which considers the deactivation of the catalyst by
Eq. (6).

g ¼ exp½aCðcÞ; ð6Þ

where η is the reduction of activity of the catalyst, C(c) is the carbonisation of

catalyst and a is the catalyst type-dependent constant. Equation (6) can be applied
to the thermo-catalytic cracking of polyethylene and polypropylene. In this case the
value of η must be calculated in each case of catalyst [10].

4.1 The Experimental Equipment and Steps

of Measurement

The goal of our research was to develop a mathematical model for our measurement
results that can be used for the determination of reaction kinetic parameters. In the
light of these data the model can become suitable to allow for the calculation of
mass-, component- as well as energy flow of the developing gas, liquid and solid
products, when processing high-volume material flow in an experimental
equipment.
By means of the well-operating model one can perform measurements on the
experimental equipment quicker and safer.

4.1.1 Steps of Experiments

We employed a tube reactor for the measurements. We started with 40 g of raw

material in each case, blending in 5 g zeolite and a nickel oxide-coated metal grid.
The experiments ran for 50 min. The raw material was infed in a batch operation in
the vertically positioned tube reactor. In the first step the raw material was dried at
105 °C for 15 min, while the entire system was being ‘flushed’ with nitrogen (the
nitrogen flow was ensured during the totality of the operation). In the second step
the temperature of the reactor was heated to 450 °C in 22 min.
During operation the hydrocarbon steams left the vertically-positioned reactor
body to the horizontal tube reactor that was kept at 300 °C. Product steams left from
this into the condenser where they were cooled down to 20 °C. Afterwards, the
liquid phase was gathered in the liquid collector, and the gas phase was flared off.
180 A. Zsemberi et al.

Table 1 Results of thermo-catalytic cracking

Products Measurement 1 Measurement 2 Measurement 3
quantity (g) quantity (g) quantity (g)
Gas fraction 10.1 9.1 11
Liquid fraction 26 27 25
Solid fraction 3.9 3.9 4
Measurement 1 Measurement 2 Measurement 3
composition (m/m%) composition (m/m%) composition (m/m%)
Gas fraction 25.25 22.75 27.5
Liquid fraction 65 67.5 62.5
Solid fraction 9.75 9.75 10

4.1.2 Experimental Results

The values of remaining gas, liquid and solid fractions obtained during measure-
ments are listed in Table 1 expressed in g.
As can be seen in the first table, the average quantities of gas, liquid as well as
solid hydrocarbon fractions produced from polystyrene and rubber waste blend
were the following: 10.1, 26, 3.9 g, which are in average mass%: 25.17, 65,
9.83 m/m%.

4.2 The Mathematical Model Fitted to the Experiments

The reaction kinetic features of the gross cracking reaction were determined by
means of the following Eqs. (4, 7–8). The activation energies of thermal cracking
processes can be calculated by means of the Arrhenius Eq. (5).

w ¼ w0
ek
t ; ð7Þ
w
ln ¼ k
t; ð8Þ
w0

where ‘w’ is the weight of the specimen, ‘w0’ is the weight of initial plastic waste,
‘k’ is the reaction rate constant, ‘n’ is the order of the reaction, and ‘t’ is the time for
cracking.
Using Eq. (8) the resulting reaction rate constants can be determined in the light
of the quantitative data of gross processes. Next, we performed further calculations
by resolving a multi-variable differential equation system according to the scheme
modelling the cracking in Fig. 4 to determine the reaction kinetic parameters of
specific partial reactions.
Utilisation of Various Hydro-Carbon-Based Wastes … 181

Fig. 4 The scheme is

modelling the cracking

Legend for the designations in Fig. 4: kg: reaction rate constant of gas formation,
kl: reaction rate constant of liquid formation, ki: reaction rate constant of interme-
diate product formation, ka: reaction rate constant of aromatic product formation,
kik: reaction rate constant of tar formation, kia: reaction rate constant of aromatic
from intermediate product, kic: reaction rate constant of chark formation. The
reaction taking place can one-by-one be described by the following differential
Eqs. (9–15).

dwM
¼ kg
wM  ki
wM  ka
wM  kl
wM ; ð9Þ
dt
dwg
¼ kg
w M ; ð10Þ
dt
dwl
¼ kl
w M ; ð11Þ
dt
dwi
¼ ki
wM  kik
wi  kia
wi  kic
wi ; ð12Þ
dt
dwa
¼ kia
wi  ka
wM ; ð13Þ
dt
dwt
¼ klk
wi ; ð14Þ
dt
dwic
¼ klc
wi ; ð15Þ
dt

where ‘wM, wl, wg, wt, wi, war, wc’ are weights of raw material, liquid phase, gas
phase, tar, intermediate products, aromatic components and chark. The ‘k’ index
represents the reaction rate constants of the mentioned components as were detailed
previously. The analytic solution of differential equation systems (9)–(15) was
performed by MATLAB software (with the Runge–Kutta solution method).
In the light of the reaction rate constants of the processes, the activation energies
were calculated by means of the Arrhenius Eq. (5).
182 A. Zsemberi et al.

Table 2 Main components Main components of rubber tyre Composition

of rubber tyre waste waste (m/m%)
Natural caoutchouc SMR5CV 29.59
Styrene butadiene caoutchouc 29.59
Carbon black ISAF N220 29.59
Stearic acid 0.59
IPPD (antiozonant) 0.89
Zinc oxide 2.96
Phenolic resin 2.37
Sulphur 0.89
CSB (accelerator) 0.89
Hexamin H-7 0.18
PVI (phthalimide) 0.12
Aromatic oil 2.34

Table 3 Values of rate Rate constants Value (1/s)

constants calculated via the
model at 450 °C ki 0.072
kg 0.022
kl 0.010
ka 4:71
109
kia 0.075
kik 0.023
kic 0.002

Table 2 presents the components of rubber tyre waste taken into consideration
when describing the mathematical model.

4.2.1 Results Obtained via the Mathematical Model

Reaction-rate constants at 450 °C obtained via the model are listed in Table 3.
Results of the reaction-rate constants indicated in Table 3 at 450 °C calculated
via the model are presented in Fig. 5.
It can be seen in Fig. 5 that the conversion of the raw material marked with light
blue curve evolves according to fidelity as it has almost fully transformed into a
product in 50 min. The quantity of gas-, liquid (aromatic fraction as well)—as well
as solid products, provided one-by-one the following: 9.2, 25.4, 5.4 g, which values
correlate with experimental results at an accuracy of 95%. Therefore, it can be
stated that the applied reaction kinetic model functions well, it is suitable for
conducting examinations at a broader temperature interval.
Utilisation of Various Hydro-Carbon-Based Wastes … 183

Fig. 5 The results calculated via the mathematical model at 450 °C

Fig. 6 The results calculated via the mathematical model at 400 °C

A relevant input parameter of the model is the temperature that has a large
impact on product distribution, thus we performed an investigation at 400 °C. We
were interested in what quantities of valuable components (mainly liquid- and
aromatic phase) develop. The results are illustrated in Fig. 6.
The results disclose well that not even the conversion of the raw material is
sufficient at 400 °C, thus, the infed solid phase converted at 93%, which appears as
a loss in the process. In addition, the combined quantity of liquid- and aromatic
phase lags behind the results measured at 450 °C as it resulted in 24.6 g. It can be
seen that it is not worth conducting experiments for parameter combinations and
raw material tested by us below 450 °C.
184 A. Zsemberi et al.

Fig. 7 The results calculated via the mathematical model at 500 °C

In the next step we performed investigations at 500 °C, whose results are
illustrated in Fig. 7.
Figure 7 reveals that the conversion of raw material is completed already in
40 min, besides, the quantity of the formed liquid- and aromatic phase added up to
27.4 g in total, which is better than the results measured and calculated at 450 °C.
Therefore, it is worth conducting further measurements also at 500 °C in order to
operate the reactor system at an optimum level as the goal is to gain the highest
possible quantity of liquid- and aromatic phases in the shortest possible time.

5 Conclusion and Determinate of Research Goals

Our work has focused on thermo-catalytic cracking technology, which is suitable

for the conversion of rubber, plastic waste and other solid hydrocarbon-based raw
material into valuable mainly liquid- and gas-state fraction. Conducting the mea-
surements and revealing the reaction kinetic parameters are complicated, costly and
not least a time-consuming task. Consequently, taking our results as a basis, we
created a general mathematical model by means of which high-safety examinations
can be performed.
The results disclose that the elaboration of the mathematical model can largely
facilitate scientific research as analyses can be conducted even without concrete
physical measurements to make optimum operational conditions more precisely.
We determined that it is not worth performing experiments at temperatures lower
than 450 °C in future. Conversely, we reached better results at 500 °C by means of
the created mathematical model. We wish to confirm the calculated results from
specific measurements in the future.
Utilisation of Various Hydro-Carbon-Based Wastes … 185

On top of that, another goal of our future research is to implement the current
model into a so-called ‘Flowsheet’ simulator, with which calculation results and
sensitivity analyses can be further refined.

References

1. Ruiz JA, Juárez MC, Morales MP, Muñoz P, Mendívil MA (2013) Biomass gasification for
electricity generation: review of current technology barriers. Renew Sustain Energy Rev
18:174–183. doi:http://dx.doi.org/10.1016/j.rser.2012.10.021
2. Stiegel GJ, Clayton SJ (2001) Gasification technologies. In: A program to deliver clean secure
and affordable energy, US Department of Energy, Office of Fossil Energy, National Energy
Technology Laboratory, 2001
3. Qi D, Zhonyang L, Kixiang Z, Jun W, Wen C, Yi Y (2012) Experimental study on bio-oil
upgrading over Pt/SO4-2/ZrO2/SBA-15 catalyst in supercritical ethanol. Fuel 103:683–692
4. Huiyan Z, Rui X, He H, Gang X (2009) Comparison of non-catalytic and catalytic fast
pyrolysis of corn cobina fluidized bedreactor. Bioresour Technol 100(3):1428–1434
5. Ying X, Tiejun W, Longlong Ma, Qi Z, Wang L (2009) Upgrading of liquid fuel from the
vacuum pyrolysis of biomass over the Mo–Ni/c-Al2O3 catalysts. Biomass Bioenergy 33
(8):1030–1036
6. Xu C, Donald J, Byambajav E, Ohtsuka Y (2010) Recent advances in catalysts for hot- gas
removal of tar and NH3 from biomass gasification. Fuel 89(8):1784–1795. doi:http://dx.doi.
org/10.1016/j.fuel.2010.02.014
7. Rasmuson A, Andersson B, Olsson L, Andersson R (2014) Mathematical modeling in
chemical engineering. Cambridge University Press; 2014
8. Szeifert F, Chován T, Nagy L (1998) System models-System analysis (Rendszermodellek-
Rendszeranalízis). Egyetemi jegyzet, Veszprém (Hungary)
9. Nikrityuk PA, Meyer B (2014) Gasification processes modeling and simulation. Wiley-VCH
Verlag GmbH & Co
10. Miskolczi N, Bartha L (2005) Investigation of thermal recycling of plastics and further
utilization of products. PhD dissertation, Veszprém (Hungary)
Development of Nitrided Selective Wave
Soldering Tool with Enhanced Lifetime
for the Automotive Industry

Zsolt Sályi, Zsolt Veres, Péter Baumli and Márton Benke

Abstract The aim of our research is to develop an economic material combination

exhibiting good wetting with solder alloy melts and enhanced lifetime against
erosion in lead-free soldering applications. Iron-nitrides have strong bonds com-
pared to metals, therefore they are expected to have increased resistance against
erosion in solder melts compared to iron soldering tools. The first step of our
research is to find the proper substrate material for nitride coating to obtain the
desired wetting behaviour. The present paper focuses on the substrate material
selection through wetting examinations of nitrided W302, 42CrMo4 and C45 steels
with SAC 305 solder alloy melt. It was found that the substrate composition
strongly affects the wetting contact angle. The best wetting was achieved with the
C45 type steel substrate. The correlation between substrate composition, com-
pounds formed during nitriding and wetting behaviour is discussed.

1 Introduction

In selective soldering applications the solder alloy melt is driven to the soldering
location with a soldering tool. In case of selective wave soldering, the tool is a
nozzle, while for hand soldering it is the tip of the soldering iron. For all cases, the
soldering tools must have good wettability with the solder alloy melt to ensure a
stable contact with the component to be soldered. In most cases, the tool is

Z. Sályi (&)
Z. Veres
P. Baumli
M. Benke

Institute of Physical Metallurgy, Metalforming and Nanotechnology,
University of Miskolc, Miskolc, Hungary
e-mail: zsolt.salyi92@gmail.com
Z. Veres
e-mail: vezso3838@gmail.com
P. Baumli
e-mail: peter.baumli@gmail.com
M. Benke
e-mail: fembenke@uni-miskolc.hu

© Springer International Publishing AG 2017 187

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_18
188 Z. Sályi et al.

produced of iron with low alloying element/impurity concentration or high purity

Armco iron. For those tools, the metallic type bonding in iron ensures the desired
wettability with the molten solder alloy. At the Fe-solder alloy melt interface, FeSn2
intermetallic layer is formed [1]. The wettability is further enhanced through the
FeSn2 intermetallic layer at the tip/solder melt interface. The FeSn2 intermetallic
layer is either produced intentionally prior to soldering (hand soldering) or during
the first soldering cycle (selective wave soldering). During application, the thick-
ness of the intermetallic layer gradually increases with soldering cycles. Fe atoms
are provided by atom transport from the tool as the FeSn2 intermetallic layer
thickens. This atom extraction appears macroscopically as the erosion of the sol-
dering tool. The erosion produces a rough surface on which ditches also form in the
later stages. The erosion is further enhanced in the ditches where the solder melt is
in contact with unharmed Fe surface. After a certain degree of erosion, the good
wetting between tool and solder melt is destroyed and the tool must be replaced.
The erosion is accelerated by the application of lead-free soldering alloys such as
SAC and INNOLOT due to the increased Sn content compared to previously used
Sn-Pb alloys and the elevated soldering temperature [2–12]. The erosion process is
more pronounced in selective wave soldering applications where the tool is exposed
to the continuous flow of solder alloy melt [9]. In such applications, the erosion
effect of the melt flow contributes to the diffusion driven Fe extraction. Finally, the
degradation of the soldering tool leads to the need to replace the soldering tool
which, in turn, requires the interruption of the soldering process, calibration of the
new tool and cause an overall loss of production time for soldering companies.
The erosion reaction of metals in Sn based solder alloys can be reduced by
composition modification of the solder alloy melt. Nishikawa et al. showed that Co
addition to Sn-Ag solder alloy effectively reduces the erosion reaction of iron plated
copper in Sn-Ag solder melt [11]. However, if the solder composition cannot be
modified, the erosion reaction must be suppressed by designing erosion resistant
solder tools. Watanabe et al. examined the erosion resistance of Fe-MWCNT (Multi
Wall Carbon Nanotube) composite in lead free solder melt. They found that the
wetting conditions decreased, but the resistance against erosion was increased [12].
Although a promising material might have been found, the fabrication of the
developed composite material is quite costly and its production can only be carried
out in laboratory conditions.
The long-term objective of our research is to develop an economic and com-
mercially obtainable substrate-coating material combination that exhibit proper
wetting with the most commonly used lead-free SAC solder alloys and increased
erosion resistance compared to the currently applied iron tools. The first step of our
research is the selection of the proper substrate material and the wetting exami-
nations of nitride coatings. For substrates, three steel types were chosen with
varying alloying contents, being W302, 42CrMo4 and C45. For coatings,
Fe-nitrides were produced at the surface of the substrates with two different
nitriding processes. In this manuscript, the results of wetting contact angle mea-
surements are presented and the correlation between contact angle and substrate
composition is discussed. The potential substrate-nitride combination of selective
Development of Nitrided Selective Wave Soldering Tool … 189

soldering tool applications is selected based on the results of the wetting exami-
nations. The examination of the erosion behaviour of the potential substrate-nitride
combination will be the next step of the research and is not yet performed.

2 Materials and Methods

Three steel types were chosen for substrates with decreasing alloying element
contents being W302, 42CrMo4 and C45, respectively. The standard compositions
of the examined steels are given in Table 1.
Plates with dimensions of 8 mm  10 mm  1 mm were cut from the bulk
alloys. The plates were austenized in an air furnace at 860 °C for 30 min, then
quenched in room temperature water. The plates were subsequently annealed at
600 °C for 20 min in air furnace. According to Böhler’s recommendation, at least
2 h of annealing is required in the case of W302 steel. The annealing time was
chosen to be shorter, since the subsequent nitriding ensured the completion of the
annealing process. After removing the decarburized surfaces the plates were divided
into two groups. The two groups were subjected to different nitriding processes. For
the first group, before the nitrocarburising a preoxidation at 350 °C for 30 min in an
air medium was carried out. Then the plates were heated up to the nitriding tem-
perature of 550 °C in nitrogen. This heating type is the standard heating process in
conventional gas nitriding that removes contaminations and produces an oxide layer
on the surfaces. After 6 h of the nitriding, the furnace was cooled down and the
samples were removed. Samples of such treatment are called here as “oxidized” For
the second group, the plates were heated up to 550 °C in nitrogen atmosphere.
Thus, no oxide layer was produced. After 6 h of nitriding the furnace cooling down
and the samples were removed. Samples of this group are called as “oxide-free”.
X-ray diffraction (XRD) phase analysis was performed with a 40 kV 40 mA
Bruker D8 Advance diffractometer using Co tube.
Before wetting angle measurements, the surfaces of the samples were cleaned
with a commercial flux, Lux-Tools DIN EN 29454. The samples were placed in an
air furnace with a small piece of SAC 305 solder alloy on the top. The composition
of the SAC 305 alloy is given in (Table 2) After holding the samples at 320 °C for

Table 1 The standard compositions of the examined steels, wt%

C Si Mn Cr Mo Ni V W Co P S
W302 0.39 1.10 0.40 5.20 1.30 – 0.95 – – – –
42CrMo4 0.38– <0.40 0.60– 0.90– 0.15– – – – – <0.035 <0.025
0.45 0.90 1.20 0.30
C45 0.42– <0.40 0.50– <0.40 <0.10 <0.40 – – – <0.045 <0.045
0.50 0.80
190 Z. Sályi et al.

20 min, they were removed and cooled down to measure the equilibrium contact
angles. If the contact angles measured on the two sides of the same sample hap-
pened to deviate with 10°, the results were neglected. Such deviations originated
from surface roughness caused by improper sample machining. The presented
results are the average of contact angles measured on the two sides of the same
sample.
Glow Discharge Optical Emission Spectrometry (GD-OES) examinations were
carried out with a GD Profiler 2 (Power: 25 W, Module: 6 V, Phase: 5 V, Pressure:
500 Pa, flushing time: 5 s, pre-integration time: 100 s).

3 Results

Figure 1 shows the X-ray diffraction patterns of the oxide-free nitride samples.
Bragg reflections of Fe2-3N (e) and Fe4N (c′) nitrides and the ferrite substrate can be
seen. No reflections of any other phases are present. Note that Co tube was applied,
which excites Cr atoms. Due to the fluorescence radiation originating from the
excitement of the Cr atoms, the sign/background ratio of W302 alloy is lower

Table 2 Composition of the SAC 305 alloy used for wetting angle measurements, wt%
Sn Ag Cu
96.34 2.95 0.59

Fig. 1 XRD patterns of the oxide-free nitride coated W302, 42CrMo4 and C45 samples
Development of Nitrided Selective Wave Soldering Tool … 191

Fig. 2 GD-OES spectra of the C45 alloy with oxide-free nitride coating

compared to 42CrMo4 and C45, therefore lower intensity peaks are not visible on
the spectrum of the W302 sample.
Figure 2 shows the near-surface region of GD-OES spectra of alloying elements
of the C45 sample with oxide-free coating. Since the equipment was not calibrated
for N and O elements, their absolute value could not be determined. However, the
distributions of these elements are well represented. For other elements, both their
distributions and absolute values could be determined. It can be seen that the
increase of the N content is monotonic with increasing depth. There is a small peak
of O in the sub-surface region, meaning that some O atoms were present during the
nitriding process. The Si and Mn show definite peaks near the surface. The relative
positions of Si, Mn and O peaks indicate that compounds/complexes of Si, Mn, and
O formed on the surface with an amount less than 2 wt%. There is a small quantity
of V in the alloys, which diffused from the bulk to the surface.
Figure 3 shows the near-surface variation of elements of the 42CrMo4 sample
with oxide-free coating. Again, the N content increases monotonically with depth in
the near surface region. Only a small peak appears in the Mn content with less than
1 wt%.
Figure 4 shows the GD-OES spectra of the W302 sample with oxide-free
coating. It can be seen that Cr, Si, Mn, O have peaks at the same positions.
Furthermore, the distribution of N is not monotonic, but having a local peak exactly
where Cr, Si, Mn and O. This means that compounds/complexes of Cr, Si, Mn, O
and N formed at the surface. The net amount of these compounds exceeds 5 wt%.
The V diffused from the bulk to the surface.
Figure 5 shows the measured equilibrium contact angles of the oxidized samples
and molten SAC 305 solder alloy systems. The W302 sample has the highest
192 Z. Sályi et al.

Fig. 3 GD-OES spectra of the 42CrMo4 alloy with oxide-free nitride coating

Fig. 4 GD-OES spectra of the W302 alloy with oxide-free nitride coating
Development of Nitrided Selective Wave Soldering Tool … 193

Fig. 5 Contact angles of the SAC305 alloy on the oxidized nitride coated W302, 42CrMo4 and
C45 substrate

Fig. 6 Contact angles of the SAC305 alloy on the oxide-free nitride coated W302, 42CrMo4 and
C45 substrate

contact angles of *40°–50°, thus, it has the poorest wetting with the SAC 305
alloy. The contact angle for the 42CrMo4 samples are *20°–30°, while those of
the C45 samples are *20°.
Figure 6 shows the contact angles of the oxide-free samples and molten SAC
305 solder alloy systems. The contact angles for the W302 samples are *70°–100°.
194 Z. Sályi et al.

As for the 42CrMo4 and C45 samples, the contact angles are the same, being
*20°. It is worth to note that the oxide-free nitriding notably increased the wetting
angles of the W302 samples compared to the oxidized nitriding. For the 42CrMo4,
there is only a slight increase in contact angles, while none is for the C45 samples.

4 Discussion

XRD examinations confirmed that Fe2-3N (e) and Fe4N (c′) nitrides were produced at
the surface of the oxide-free nitrided samples. For the C45 and 42CrMo4 alloys,
GD-OES examinations revealed that no notable quantity of oxides/nitrides of alloying
elements or impurities formed during the oxide-free nitriding. The formation of such
compounds was inhibited because of the low alloying element concentration of the
C45 and 42CrMo4 steels. It was concluded that the good wetting of SAC 305 sol-
dering alloy melt on the oxide-free nitrided C45 and 42CrMo4 samples were due to the
lack of additional oxides/nitrides, i.e. the desired wetting originated from the
favourable adhesion between the Fe2-3N (e) and Fe4N (c′) nitrides and the SAC 305
solder alloy melt. As for the W302 alloy, high wetting angles were measured after the
oxidized nitriding, furthermore, wetting angles were even higher after the oxide-free
nitriding. According to GD-OES examinations, nitrides of alloying elements such as
Cr, Si and Mn formed during the oxide-free nitriding at the surface of the W302 alloy.
It is well known that CrN is non-wettable for most metal melts, and, because of this
character, it is used as non-wetting coating in soldering and aluminum casting pro-
cesses [13–15]. It was deduced that the poor wetting of molten SAC 305 alloy on the
nitrided W302 alloy was due to the presence of non-wetting compounds on the surface
during nitriding. Finally, the obtained results showed that low alloyed steel substrates
such as C45 or 42CrMo4 with nitride coatings can be candidates for wetting coatings
within lead-free soldering applications.

5 Conclusions

The suitability of steel substrate-nitride coating combinations as wettable materials

with enhanced erosion resistance for lead-free soldering applications are being
investigated. This paper reports about the correlation between substrate composi-
tion and wetting behaviour. Samples of W302, 42CrMo4 and C45 steels were
covered with Fe2-3N (e) and Fe4N (c′) nitride layers using gas-nitriding. For the first
set of alloys, conventional oxidizing was performed prior to nitriding, while for the
second sets, the oxidizing step was skipped. The wettability was examined in the
pre-oxidized and oxide-free samples/SAC 305 solder alloy systems. It was found
that the composition of substrates strongly affects the wettability through formation
of nitrides and oxides of alloying elements at the surface. The best wetting with
contact angles of *20° was achieved with the C45 type steel substrate. It was
Development of Nitrided Selective Wave Soldering Tool … 195

deduced that low alloyed steel substrates with nitride coatings can be candidates for
wettable materials for lead-free soldering tools.

Acknowledgements The authors are grateful for Tibor Kulcsar for the GD-OES examinations.
The described article was carried out as part of the EFOP-3.6.1-16-00011 “Younger and Renewing
University—Innovative Knowledge City—institutional development of the University of Miskolc
aiming at intelligent specialisation” project implemented in the framework of the Szechenyi 2020
program. The realization of this project is supported by the European Union, co-financed by the
European Social Fund. Marton Benke was further supported by the Postdoctral Researcher
Fellowship of the Hungarian Academy of Sciences and Peter Baumli by the Janos Bolyai Research
Fellowship of the Hungarian Academy of Sciences.

References

1. Nishikawa H, Takemoto T, Kifune K, Uetani T, Sekimori N (2004) Effect of iron plating

conditions on reaction in molten lead-free solder. Mater Trans 45(3):741–746. doi:10.2320/
matertrans.45.741
2. Information on http://www.elexp.com/Images/Weller_Coping_with_Lead_Free.pdf
3. Shangguan D (2005) Lead-free solder interconnect reliability. ASM International.
ISBN:978-0-87170-816-8
4. Henshall G, Bath J, Handwerker CA (2011) Lead-free solder process development. Wiley.
ISBN:978-0-470-41074-5
5. Ipser H (2007) European cost action 531: basic research on lead-free soldering. J Mining
Metallurgy 43B(2):109–112. doi:10.2298/JMMB0702109I
6. Kroupa A, Dinsdale AT, Watson A, Vrestal J, Zemanova A (2007) COST531 Project—study
of the advanced materials for lead-free soldering. J Mining Metallurgy 43B(2):113–123.
doi:10.2298/JMMB0702113K
7. Kroupa A, Dinsdale A, Watson A, Vřešťál JJ, Zemanova A, Broz P (2012) The
thermodynamic database cost MP0602 for materials for high-temperature lead-free soldering.
J Mining Metallurgy 48(3)B:339–346. doi:10.2298/JMMB120711043K
8. Takemoto T, Uetani T, Yamazaki M (2004) Dissolution rates of iron plating on soldering iron
tips in molten lead-free solders. Soldering Surface Mount Technol 16(3):9–15. doi:10.1108/
09540910410562473
9. Nishikawa H, Kang S, Takemoto T (2009) Erosion behavior and interfacial reaction of
stainless steels in molten lead-free solder. Trans JWRI 38(2):53–56
10. Information on http://www.circuitspecialists.com/blog/soldering-tip-care-tips
11. Nishikawa H, Komatsu A, Takemoto T (2005) Interfacial reaction between Sn–Ag–Co solder
and metals. Mater Trans 46:2394–2399. doi:10.2320/matertrans.46.2394
12. Watanabe J, Sekimori N, Hatsuzawa K, Uetani T, Shohji I (2010) Study on erosion resistance
characteristics of Fe-MWCNT composite plating with respect to lead-free solder. J Phys: Conf
Ser 379(012025):1–10. doi:10.1088/1742-6596/379/1/012025
13. Wang Y (1997) A study of PVD coatings and die materials for extended die-casting die life.
Surf Coat Technol 94–95:60–63. doi:10.1016/S0257-8972(97)00476-3
14. Gulizia S, Jahedi MZ, Doyle ED (2001) Performance evaluation of PVD coatings for high
pressure die casting. Surf Coat Technol 140:200–205. doi:10.1016/S0257-8972(01)01037-4
15. Panjan P, Cekada M, Kirn R, Sokovic M (2004) Improvement of die-casting tools with
duplex treatment. Surf Coat Technol 180–181:561–565. doi:10.1016/j.surfcoat.2003.10.119
The Effect of Tensile Strength
on the Formability Parameters of Dual
Phase Steels

Gábor Béres and Miklós Tisza

Abstract The importance of dual phase (DP) steels in the automotive industry was
continuously spreading in the last decade. With their special microstructure—
containing ferrite and martensite in particular ratio—high strength and increased
formability is available. That’s the reason why the application of DP steels is
providing to exceed the 50% in a modern car body structure, according to the
European program of Ultra-Light Steel Auto Body Advanced Vehicle Technology.
This paper presents the experimental results of hemispherical dome tests and uni-
axial tensile tests of three types of DP steels: DP 600, DP 800 and DP 1000. The
effect of the tensile strength on the formability was investigated. It was described by
the total and the ultimate tensile elongation, the average anisotropy and the limiting
dome height (LDH). Based on our results it can be concluded that both the total and
the uniform elongations are nearly linearly decreasing as the tensile strength is
increasing. The slope of total elongation is more sensitive to the strength growth.
However, it is no longer true for the plastic anisotropy. The reduction rate of
average anisotropy stops over 800 MPa, and does not change until 1100 MPa.
According to the dome tests results, the formability is also influenced by the sample
geometry—through the deformation path—besides the tensile strength. The LDH
values in biaxial stretch strain conditions are less dependent on the tensile strength.
They are within a 3 mm interval for all three strength classes. In plane or
stretch-press strain conditions, higher reduction can be observed. The characteris-
tics of dome height curves are similar for all samples, regardless of their strength.

G. Béres (&)
Department of Materials Technology, Faculty of GAMF Technical and IT,
University of Pallasz Athéné, Kecskemét, Hungary
e-mail: beres.gabor@gamf.kefo.hu
M. Tisza
Department of Mechanical Technologies, Faculty of Mechanical Engineering and IT,
University of Miskolc, Miskolc, Hungary
e-mail: tisza.miklos@uni-miskolc.hu

© Springer International Publishing AG 2017 197

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_19
198 G. Béres and M. Tisza

1 Introduction

Weight-loss developments in the vehicle industry resulted in relevant research results

in the science of structural materials. Spreading of non-ferrous materials—like alu-
minium alloys and composites—can be observed, but the steels still remain deter-
minative materials for car body manufacturing. To reducing weight, newly developed,
advanced high strength steels (AHSS) and ultrahigh strength steels (UHSS) have been
used as car body elements. Earlier [1] and present-day results [2]—validated with FE
modelling simulations—showed that using these steels allows thinner components for
lightweight constructions with the same, or even with improved stiffness and crash-
worthiness. Furthermore, a significant reduction of fuel consumption and greenhouse
gas emission can be available in this way. 57 kg weight reduction is equivalent to
0.09–0.21 l/km fuel saving [2].
Application of AHSS materials significantly increased formability beside higher
strength compared to conventional steel types. AHSS includes dual phase (DP),
complex phase (CP) and transformation induced plasticity (TRIP) steels [3].
DP steels are typically multi-phase steels, widely used in the automotive industry
due to their favourable high strength and good formability. The application of these
multi-phase steels for car body panels production exceeds 50% according to the
European ULSAB-AVC (Ultra Light Steel Auto Body—Advanced Vehicle
Concept) consortium in 2011 [4].
DP steels are commonly manufactured by hot rolling, which followed by fast
cooling from the inter-critical temperature interval below to Ms temperature. In this
way, their microstructure consists of hard martensite islands in soft ferrite matrix.
The mechanical properties of DP steels are mainly depending on the amount, the
morphology, the dispersion and the cohesion of these two phases [3–6]. With key
factors of the manufacturing, the characteristics of the ferrite and martensite par-
ticles can be significantly controlled. These key factors are the holding temperature
or the holding time after the hot rolling, and the cooling rate respectively. Many
studies [4–7] present the effect of the grain structure for the formability of DP steels.
Some of them [4–6] use microstructure-based modelling, some other work [7] deals
with experimental investigations. Microstructure-based models use representative
volume elements (RVEs) to describe the multiphase material features. The defini-
tion of such model is time-consuming, however can lead to a more precise solution.
This paper is intended to present the experimental results of the effect of the
tensile strength on the formability parameters of DP steels. Measuring the tensile
strength is easier than that of the microstructure properties and provides sufficiently
accurate benchmark for industrial users.
The Effect of Tensile Strength on the Formability Parameters … 199

Table 1 Chemical composition of the investigated DP steels (%)

C Si Mn P S Cr
DP 600 0.085 0.171 0.872 0.013 0.005 0.014
DP 800 0.161 0.187 1.520 0.012 0.003 0.027
DP 1000 0.148 0.181 1.500 0.012 0.004 0.042

Fig. 1 Modified Nakajima specimens

2 Samples and Materials

Three types of commercially available dual phase steels—DP 600, DP 800 and DP
1000—were investigated by mechanical testing. Initial table sizes were
2000
1250
1 mm. The chemical compound of the applied materials is sum-
marized in Table 1.
The given values were acquired by optical spectroscopy measurements. The
results are in good agreement with the supplier’s certification [8]. It is worthwhile to
note that DP 800 has slightly higher carbon content than DP 1000. Nevertheless,
both types fall into the manufacturing tolerances.
Tensile tests and limiting dome height (LDH) tests were applied as mechanical
investigations. Tensile tests were carried out in accordance with MSZ EN ISO 6892
standard. The crosshead speed was 25 mm/min. Both the tensile and doming
samples were cut by laser beam from 1 mm thick blanks, with tensile sample
orientation 0°, 45° and 90° angles with respect to the rolling direction. Their
dimensions also aligned to the standard mentioned above. Doming specimens
called modified Nakajima specimens (Fig. 1) were manufactured based on literature
[9]. The longitudinal axes of the samples were perpendicular to the rolling direc-
tion, since it is known that this direction shows the worst results in terms of stretch
formability [9]. The applied blank holder force was 120 kN, which totally prevent
the movement of the sheets under the blank holder. The diameter of hemispherical
punch was 100 mm and it passed with 50 mm/min stroke speed. Surface conditions
were prepared without lubrication.
200 G. Béres and M. Tisza

3 Results

3.1 Tensile Tests

Primary values of tensile test results can be seen in Table 2. The engineering
stress-strain curves are shown in Fig. 2. It can be seen, that the ultimate tensile
strength values are close to the values of denomination signs of each steel grade.
The highest tensile strength is measured at the DP 1000 and then it continuously
decreases toward the DP 600. Conversely, the total elongation increases with the
strength reduction.
Plastic anisotropy (r ) cannot be deduced from the figure, but reports the thinning
tendency of the materials. The calculation method of it can be described by Eq. (1):

r0 þ r90 þ 2  r45
r ¼ : ð1Þ
4

In this context r0, r90 and r45 are the measured values in the direction 0°, 90° and
45° compared to the rolling direction. The relationship between each r values in
different directions can be followed on Fig. 3. It is observed that in case of the DP
600 the anisotropy increases linearly and the highest r value appears at 90°. The

Table 2 Basic mechanical parameters

E (N/mm2) Re (N/mm2) Rm (N/mm2) A80 (%) r
5
DP 600 2.1*10 445 656 20.6 0.92
DP 800 2.1*105 571 879 16.0 0.75
DP 1000 2.1*105 767 1099 10.6 0.76

Fig. 2 Engineering 1200

stress-strain curves
1000 DP600
engineering stress (MPa)

DP800
800
DP1000
600

400

200

0
0 0.05 0.1 0.15 0.2 0.25
engineering strain (%)
The Effect of Tensile Strength on the Formability Parameters … 201

DP600 DP800 DP1000

1.10
1.00
plastic anisotropy-r 0.90
0.80
0.70
0.60
0.50
0.40
0 45 90
direction compared to the rolling angle (°)

Fig. 3 Plastic anisotropy values in each specimen direction

Fig. 4 The average 1.00

anisotropy dependence on
average anisotropy

tensile strength 0.90

0.80
0.70
0.60
0.50
0.40
600 700 800 900 1000 1100 1200
tensile strength -Rm (MPa)

functions’ forms are not linear and the changing tendency shows opposite beha-
viour for the DP 800 and DP 1000.
The dependence of the average anisotropy on the tensile strength is added by
Fig. 4. Based on the tensile strength—which is a function of the microstructure
properties, like martensite volume fraction, martensite carbon content or grain size
—the average anisotropy does not significantly change over 880 MPa. It increases
below that, up to 660 MPa surely. This fact means, that such DP steels could
behave similarly during deep-drawing operation, which has higher strength than
880 MPa. It is also confirmed by our previous research work [10].
The changing attitude of the total and uniform elongations deviate from the
anisotropy in the function of tensile strength. Both parameters are decreasing lin-
early, as the tensile strength increases (Fig. 5). From these results, it can be con-
cluded that the stretch formability of such steels also decrease linearly with the
increase of strength.
202 G. Béres and M. Tisza

total elongation uniform elongation

25.0

20.0 y = -0.0226x + 35.534

elongation (%)
R² = 0.9981
15.0

10.0
y = -0.0136x + 21.896
5.0
R² = 0.992
0.0
600 700 800 900 1000 1100 1200
tensile strength - Rm (MPa)

Fig. 5 Changing of the elongation in the function of strength

Table 3 LDH values at Bridge-width 20 40 80 125 200

different specimen (mm)
bridge-widths
DP 600 21.3 27.3 24.9 26.9 37.2
DP 800 19.3 23.9 23.9 24.4 37.4
DP 1000 15.1 20.0 20.2 22.1 35.1

3.2 Limiting Dome Height Tests

Limiting dome height (LDH) tests were performed to measure the maximal dome
height for different sample geometries. The applied geometries (shown in Fig. 1)
make possible the investigation of the effect of deformation path on stretch
formability. The measured limiting dome height values in mm, and the applied
sample bridge-width dimensions are shown in Table 3.
The LDH curves in the function of bridge-width are given by Fig. 6. Note that
the LDH values are decreasing with the increase of the strength. The characteristics
are similar, the lowest is at 20 mm width, and the highest is at 200 mm width for all
materials. From these results it can be concluded that these steels resist better
against stress in biaxial stretch strain condition than in-plane or press-stretch strain
conditions.
If the LDH values are taken into account as the function of the tensile strength,
the biaxial stretch strain condition seems to provide the best formability likewise
(Fig. 7). Samples marked by “LDH_200” refers to the LDH of the specimens with
200 mm bridge-width, but different strength. It can be stated that the changing
characteristic is nearly linear, and the dome height results are within 3 mm interval,
while the tensile strength is almost doubled. Higher changing—roughly 5 mm—
appeared in the already lower LDHs in plane strain condition. It is displayed by
“LDH_80” nomination in the figure. The biggest deviation is discovered in
The Effect of Tensile Strength on the Formability Parameters … 203

40
DP600 DP800 DP1000
35
LDH (mm) 30

25

20

15

10
20 40 60 80 100 120 140 160 180 200
specimen bridge-width (mm)

Fig. 6 LDHs at different sample geometry

40 LDH_200
35 y = -0.0046x + 40.602 LDH_80
R² = 0.6642
30
LDH (mm)

LDH_20
25 y = -0.0106x + 32.322
R² = 0.898
20

15 y = -0.0139x + 30.784
R² = 0.957
10
600 700 800 900 1000 1100 1200
tensile strength - Rm (MPa)

Fig. 7 The effect of the tensile strength and deformation path for the stretch formability

compress-stretch deforming path (LDH_20). Thus the LDHs less dependent on the
tensile strength in biaxial stretch processes, but its influencing effect intensifies with
the approaching of the compressing deformation states.

4 Summary

Mechanical investigations of three types of dual phase steels were performed by

tensile and limiting dome height tests. The exact material types were commercially
available DP 600, DP 800 and DP 1000. Since the strength properties are in direct
connection with their microstructure, formability parameters were investigated
according to their tensile strength as the basis of analysis. Total and ultimate
elongation, anisotropy and limiting dome height behaviour were also examined.
From these investigations the following main conclusions may be drawn:
204 G. Béres and M. Tisza

• The changing characteristic of plastic anisotropy in different deforming direc-

tions is similar for DP 800 and DP 1000. Average anisotropy doesn’t decrease
further over 800 MPa.
• Total and uniform elongation reduces linearly with the increase of the strength.
The slope of the total elongation reacts stronger for the strength modification.
• LDH values are significantly influenced by both the tensile strength and the
deformation paths. Favourable deformation capability can be observed in biaxial
stretch strain conditions, almost independently the materials’ strength. In plane
and compress conditions, lower LDH values and stronger strength dependence
can be observed.

Acknowledgements The presented research work was supported by Ministry of Human

Resources, in the program of National Talent Program NFTÖ-16—scholarships to provide indi-
vidual development.

References

1. Li Y, Lin Z, Jiang A, Chen G (2003) Use of high strength steel sheet for lightweight and
crashworthy car body. Mater Des 24:177–182
2. Cui X, Zhang H, Wang S, Zhang L, Ko J (2011) Design of lightweight multi-material
automotive bodies using new material performance indices of thin-walled beams for the
material selection with crashworthiness consideration. Mater Des 32:815–821
3. Kuziak R, Kawalla R, Waengler S (2008) Advanced high strength steels for automotive
industry. Arch Civil Mech Eng VIII:103–117
4. Uthaisangsuk V, Prahl U, Bleck W (2011) Modelling of damage and failure in multiphase
high strength DP and TRIP steels. Eng Fract Mech 78:469–486
5. Surajit Kumar Paul (2013) Real microstructure based micromechanical model to simulate
microstructural level deformation behavior and failure initiation in DP 590 steel. Mater Des
44:397–406
6. Sun X, Choi KS, Soulami A, Liu WN, Khaleel MA (2009) On key factors influencing ductile
fractures of dual phase (DP) steels. Mater Sci Eng, A 526:140–419
7. Movahed P, Kolahgar S, Marashi SPH, Pouranvari M, Parvin N (2009) The effect of
intercritical heat treatment temperature on the tensile properties and work hardening behavior
of ferrite–martensite dual phase steel sheets. Mater Sci Eng, A 518:1–6
8. Official certificate of material testing laboratory of SSAB EMEA AB, Borlange, Sweden
9. Tisza M, Kovács PZ, Lukács Zs (2015) Formability of high strength sheet metals with special
regard to the effect of the influential factors on the forming limit diagrams. Mater Sci Forum
812:271–275. doi:10.4028/www.scientific.net/MSF.812.271
10. Danyi J, Végvári F, Béres G (2016) Járműipari célú acéllemezek mélyíthetőségi
és mélyhúzhatósági problémái. Miskolci Egyetem Közleményei: Anyagmérnöki Tudományok
39(1):19–28
Comparison of Two Laser Interferometric
Methods for the Study of Vibrations

Miklós Béres and Béla Paripás

Abstract We performed a comparison between two laser interferometric methods.

One of the methods is used by our modernized Michelson type laser interferometric
motion analyser (LIMA), which measures primarily the displacement, while the
other method is the one used by an LDV (Laser Doppler Vibrometer) device, which
primarily measures the velocity on the basis of the Doppler effect. The motion that
is measured is the vibration of the corner cube reflector (retroreflector) of the
measuring arm of the LIMA. During the measurements the LDV measured the back
plate of the house of the retroreflector. These simultaneous analyses of the vibra-
tions made possible the comparison of these two methods for the precise velocity
studies. In general, it was found that the velocity values, obtained by the two
methods correspond to each other within 2% accuracy.

1 Introduction

Laser measurement techniques, especially the laser interferometrical methods, are

widely used in automotive developments [1–3]. The Institute of Physics at the
University of Miskolc and the Wigner Research Centre for Physics developed a
precision laser interferometric motion analyser (LIMA) [4, 5] more than two decades
ago. At the present, it serves demonstration purposes, but the goal is to make it
appropriate for the completion of practical tasks (e.g. certifying industrial equipment).
Since the construction was relatively old, the technical part had to be converted in
accordance with the requirements of the present day. Some operating units of the
equipment remained—such as the laser, optics, detectors, etc., but the control of the
analyser and the analysis of the measured data had to be upgraded. Due to the oper-
ating principle of the device—which we describe later—it measures the displacement

M. Béres (&)
B. Paripás

University of Miskolc, Miskolc, Hungary
e-mail: mechbere@uni-miskolc.hu
B. Paripás
e-mail: fizpari@uni-miskolc.hu

© Springer International Publishing AG 2017 205

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_20
206 M. Béres and B. Paripás

accurately even without calibration, and the velocity of the point can be calculated
accurately from the sub-micron resolution data (with the appropriate software back-
ground). The tests and developments were carried out in the Institute of Physics [6]
with the support of the Wigner Research Centre for Physics.
Our other motion analyser that uses the interferometric principle is a Polytech
PDV-100 type LDV (Laser Doppler Vibrometer) device [7]. This device primarily
measures the velocity, therefore, based on its operating principle it fundamentally
differs from the displacement meter we developed. Measuring simultaneously with
the two precision devices that use different principles, comparing the measurement
results, and analysing the differences is certainly an exciting engineering task. This
measurement comparison can be considered as the calibration of the LDV device,
which is thought to be less precise.

2 Interferometric Measuring Systems

2.1 The Laser Interferometric Motion Analyser (LIMA)

The device is essentially a Michelson interferometer in which, by exploiting the

polarisation, we doubled the accuracy and made the direction of motion also
measurable. The increased accuracy is k/8. Numerous such solutions can be found
in the literature [8–10].
The opto-mechanical drawing of the interferometer is shown in Fig. 1. The heart
of the interferometer is a temperature-stabilised single-mode He–Ne laser
(UNIPHASE-1007). The frequency of its mode is 4.7375
1014 Hz, which corre-
sponds to a wavelength of 632.82 nm in vacuum and 632.64 nm in normal air. The
laser beam formed by the beam expander is split into two perpendicular, approx-
imately the same intensity beams by the polarizing beam splitter prism in such a
way that the polarisation of the two beams are also perpendicular to each other. One
beam is reflected from the reference corner cube prism (at the top of the diagram),
while the other is reflected from the movable measuring corner cube prism (on the
right side of the diagram). The beams meet again in the same polarizing beam
splitter prism where they previously split up. The phase relationship of the two
beams is determined by the path length difference between the two arms [4].
Naturally, this path length difference has to be within the range of the coherence
length of the laser (which can be a few meters).
The combined beam is led to the detector unit by mirrors. There the laser beam is
split into three parts by beam splitter prisms. After creating the appropriate phase
shifts (0, k/2, k/4), these are detected by three independent photodiodes (BPx61)
behind polarizers with the appropriate orientation. Out of the signals of these
photodiodes, A and B are in opposite phase, while between A and C there is a 90°
phase difference. Therefore, in total the detector system detects a maximum (or
minimum) at every 90° (quarter period).
Comparison of Two Laser Interferometric Methods … 207

Fig. 1 Opto-mechanical outline of the interferometer (LIMA) at the institute

In this device the measurement is based on the detection of these, therefore the
achievable accuracy in terms of change in the path length of light is k/4. This in turn
corresponds to a displacement of k/8 for the corner prism in the measuring arm
(since the light travels there and back). The signal order of the three detectors also
determines the direction of motion of the measuring arm.
A measuring card and a LabView program written for the measurement are
connected to the measuring system. The signals are transferred from the detector
unit to the central unit. This requires an appropriate measuring card that can handle
the signals of the three detectors (which in principle can follow a motion of 15 cm/s
in the measuring arm, corresponding to 3 million pulses) and can also determine the
signal order of the detectors. The National Instruments NI USB-6341 OEM type
measuring card is connected to the computer (laptop) through a USB port. On the
card the operation of the analogue, digital and counter subsystems are coordinated
by the NI-STC3 timing and synchronisation technology, providing independent
timers for the analogue and digital I/O subsystems that are on the same card. For us
the four improved 32 bit counters are extremely important, which we can use for
frequency measurement, pulse width measurement, and encoder operations. This
latter is necessary for determining the signal order of the detectors, i.e. the direction
of motion. In principle the counters can count at a speed of 100 MHz, which can be
read by the computer at a frequency of 1 MHz, therefore the data handling and data
transfer speed of the card is more than enough for our photo detectors.
The numerical and graphical visualisation of the metering data is also performed
by the measuring program (Fig. 2). This system can be used to study various
movements that we can generate for example by a differential screw micrometer, or
by gently pushing and hitting the vibration-free steel table. The study of these is
aided by the built-in Fourier analyser (FFT) that shows the frequency spectrum of
the created vibrations.
208 M. Béres and B. Paripás

Fig. 2 The continuous measurement mode of the measuring program

2.2 The LDV (Laser Doppler Vibrometer)

The LDV (Laser Doppler Vibrometer) is a device that is suitable for contactless
measurement, which we can use to examine the vibration of various objects. The
laser beam exiting the LDV equipment has to be focused onto a surface of the
object to be examined, and the velocity function of this surface can be inferred
using the frequency of the reflected laser light based on the Doppler-effect [10]. If
we are detecting the wave that is reflected from an object moving at velocity v (in
case of a stationary wave source), then the detected frequency f differs from the
original f0 frequency by an fD Doppler shift:

cþv  v
f ¼ f0 ffi f0 1 þ 2 ¼ f0 þ fD ð1Þ
cv c

Therefore
v v
fD ¼ 2f0 ¼ 2 ð2Þ
c k

In case of a receding reflecting object the signs are opposite; c is the phase
velocity of the wave in the given medium. Naturally, this is only true if there is no
cosine error, i.e. if the laser beam and the velocity vector are parallel.
Figure 3 shows the schematic setup of the vibrometer. The laser beam arriving
from the helium–neon (He–Ne) laser (with frequency f0) is split by the beam splitter
into a reference and a measuring beam. The measuring beam arrives onto the target
Comparison of Two Laser Interferometric Methods … 209

Fig. 3 Schematic setup of the vibrometer

object, from which it reflects and changes its frequency by an fD Doppler shift
because of the motion of the object in accordance with Eq. 2.
The light is reflected from the target object in every direction, but a portion of the
light is collected by the LDV optics, and it is reflected onto the photo detector
through the beam splitter. The frequency of this light is f0 + fD. The reference beam
passes through the Bragg cell, which adds a frequency shift of fB (in our case
20 MHz). The scattered light interferes with the reference beam on the photo
detector. The frequency difference of the two interfering waves also appears
(fB − fD), and this value falls into the 10 MHz range. This is such an intensity
variation that the photo detector can already detect (it cannot follow the original
1014 Hz variation). The output of the photo detector is a standard frequency
modulated (FM) signal, with the Bragg cell as carrier frequency and the Doppler
shift as modulation frequency. From this signal the time dependence of the velocity
of the target object can be determined by demodulation. The LDV equipment we
use is a Polytec PDV-100 (Portable Digital Vibrometer) type vibrometer.

3 Measurements Results

In the previous chapters we showed that the LIMA is able to analyse the motion (its
component that is in the direction of the laser beam) of the corner prism in the
measuring arm. The LDV on the other hand does not require a corner cube prism
(that is fixed onto the point to be measured), although a reflective film dot that is
stuck onto the point greatly helps the measurement. Therefore, both devices are able
to analyse the vibrations of the gently knocked corner prism in the measuring arm
of the LIMA. The simplest way for this is when the two lasers of the two devices
shine onto the two opposite sides of the corner prism. Thus the devices measure the
projection of the motion along a common line, but from opposite directions.
210 M. Béres and B. Paripás

Consequently, if the corner cube prism is receding from the laser of the LIMA, then
it is approaching the laser of the LDV. It should be noted that the velocities for the
two instruments still have the same sign, because the LIMA considers moving away
to be a positive displacement, while for the LDV approach has positive velocity.
We performed our earlier comparison measurements on the old “vibration-free”
table of the laser interferometry laboratory of the Institute of Physics, on the same
table that the laser interferometric motion analyser was originally assembled. The
quotation marks are warranted because the table mainly protected the analyser from
the vibrations of the floor, however, it mutually transmitted the vibrations of the
analyser units (laser, beam splitter prism, measuring prism) mounted to different
points on the table. The situation was also worsened by the fact that we could only
place the LDV equipment on a stand (tripod) that was independent from the table.
All these resulted complex, hardly analysable vibration images that were slightly
different on the two devices, which made the calibration unreliable as well.
During the last few months we managed to acquire a vibration-free Nexus
Breadboard table top with a size of 900  1800  110 mm, which we placed onto
our custom built frame and air springs. The vibration-free table configured this way,
in addition to easily decreasing the amplitude of the vibrations of the environment
below the measurement threshold of our laser interferometric motion analyser
(0.1 µm), it effectively separates the units that are mounted at different points on the
table. Therefore on this table the vibrations of e.g. the gently knocked corner prism
in the measuring arm does not spread onto e.g. the beam splitter prism. With the
newly acquired units we could also solve the mounting of the LDV onto the table.
The measurements presented in this paper were already carried out on this table
(Fig. 4).
As we have indicated in the preceding chapters, the LIMA primarily measures
displacement, and the LDV measures velocity. Therefore the obtained graphs
cannot be compared directly, e.g. from the displacement data we have to derive the

Fig. 4 Positioning of the devices

Comparison of Two Laser Interferometric Methods … 211

velocity. This is done by the LabView program with the frequency that can be set in
its graphical interface (it divides the displacement measured in the previous time
interval with the length of the interval). For these measurements we typically set a
frequency of 1200 Hz (T = 833.33 ls), because the 1200 sample/s (and its multi-
ples) can also be chosen on the LDV. If we want to compare the two velocity
graphs, we also have to match the timescale of the two graphs. Since both mea-
suring programs are run by the same machine, the factor of the two timescales are
matching in principle, therefore the time differences should also match precisely.
However, this is still not the case, in a 10 s measurement interval, the LDV velocity
graph shifts by approximately 1 ms (one or two channels) relative to the one of the
LIMA. We could not figure out the reason for this, initially we considered it as an
error to be eliminated, but later it even helped our work. Namely, it helped with
aligning the starting values of the scales. This was needed because we could not
arrange for the ls precision simultaneous start of the two types of measurements.
The rough aligning (that is about 0.5 ms precision) of the starting values of the two
graphs can be done manually, because the knocking of the corner prism in the
measuring arm starts the vibrations in a well determined way. The best match
(possibly the best two matches) can be found visually, then our inaccuracy is
therefore T/2 (0.5 ms) at most. The further increase of precision is helped by the
10-4 relative deviation of the timescales, because due to this there will surely be
such a knock, where the alignment is more precise by an order of magnitude. This
knock will be the subject of our comparative studies. Finding the best alignment
was also helped by the time graph of the deviation of the velocity values measured
by the two methods.
Figure 5 shows the result of a 12 s long knocking sequence, that is the damped
vibrations following the knocks, on the aligned timescale. In the upper part of the
figure the LDV (black) and LIMA (red) velocity data are represented together. (The
LDV data are corrected, see later.) The two data sequences completely overlap, the
differences cannot really be seen at this magnification.
Therefore, we calculated the differences as well, and are showing those at the
bottom part of the figure. They show it quite well that the largest discrepancies are
always at the first oscillation following the knock. Moreover, it can also be seen that
after this the differences are smaller, but they have approximately constant ampli-
tude all the way until the vibrations completely die down.
This is especially true after the knocks at around the 4th and the 7th second,
where the alignment is obviously the best. In these ranges the discrepancies
(therefore independently of the amplitude of the vibration, and excluding the first
oscillation) steadily fall into the region between −0.1 and +0.1 mm/s. The reason
for this definite maximum can be easily understood. Namely, the measurement
precision of the LIMA (as we indicated earlier) is k/8  0.08 lm, and the length of
a time window is 1/1200 s. The ratio of these two gives 96 lm/s  0.1 mm/s,
therefore this is the velocity measurement precision of the LIMA (the random error
in the applied measuring program and sampling frequency). This velocity mea-
surement precision could be improved by decreasing the sampling frequency, but
then the equipment would be less sensitive to the faster vibration components.
212 M. Béres and B. Paripás

Fig. 5 The velocity data of the damped vibrations following a knocking sequence: upper LDV
(black) and LIMA (red) velocity data together, lower their difference

Fig. 6 The first few oscillations after the 8th knock zoomed in (The data are as in the previous
diagram.)

In Fig. 6 we can also look at the few oscillations after the knock that happened at
6.97 s (otherwise the eighth) zoomed in. Starting from the second period, both
devices measure a damped vibration of 67.8 Hz, with an almost perfect alignment.
The significant difference is in the first half period. There the LDV measures an
approximately 260 Hz quickly decaying component more definitely than the LIMA.
Comparison of Two Laser Interferometric Methods … 213

Fig. 7 The cessation of the oscillations after the 8th knock (The data are as in the upper part of
Fig. 6.)

It’s definitely because the slow vibration is the vibration of the whole corner prism,
while the fast vibration is the vibration of the elements of the prism relative to each
other. The light of the LDV laser is reflected at the back and in one point, while that
of the LIMA is reflected at the front and in three points (on the three sides of the
prism one after another). Therefore the LIMA somewhat averages the fast
vibrations.
The monitoring of the dying down of the vibrations is also instructive. Figure 7
is essentially the continuation of Fig. 6 (with the omission of a few periods, and on
a finer velocity scale). It can be seen that the LDV nicely follows the oscillations the
whole time, which are not purely sinusoidal towards the end; they consist of the
sum of multiple vibrations. However, in case of the LIMA the velocity becomes
very spiky, essentially only three velocity values occur: zero and ±96 lm/s,
depending on the fact if there was a signal arriving in the previous time window,
and which sign displacement it belonged to (see above). With the decrease of the
amplitude, the values that are different from zero become more and more scarce,
and eventually they completely disappear. According to what has been written
earlier, the LIMA cannot yet see the vibrations with amplitudes less than 50 nm. In
the case of 67.8 Hz this measurement threshold corresponds to a (v = Ax) 20 lm/s
velocity measurement threshold, and this is also confirmed by the diagram.
Based on the foregoing, we can clearly state that with the current measuring
program the LIMA is less suitable for the study of movements with a velocity of
0.1 mm/s than the LDV. This is so because although the LIMA measures the
displacement very accurately (in k/4 units), it derives the instantaneous velocity
(and acceleration) inaccurately from this, because the time windows are not syn-
chronised to the arrival of the signals. Therefore, this derivation method (which we
have already described at the beginning of the chapter) should be developed by all
means. We emphasize that the “inaccuracy” of the LIMA applies only to the
instantaneous velocity values; it does not apply to the average velocities
214 M. Béres and B. Paripás

Fig. 8 Determination of the

optimal LDV correction
factor

Table 1 The LDV correction factors corresponding to the measurement points on the corner cube
prism
Point number Coordinates (mm) Correction factor LDV excess (%)
1 (0;0) 0.98 2.0
2 (0;7) 0.91 9.9
3 (0;14) 0.86 16.3
4 (0;−7) 1.05 −4.8
5 (−14;0) 1.00 0
6 (14;0) 1.00 0

corresponding to longer time periods. Those are very precise, because we calculate
them from high precision data.
The key objective of the comparison measurements was to determine if the two
measurement methods provide the same measurement results, and if there is a
possibility for making the calibration of the LDV, which is thought to have a less
precise scale, more precise. Figure 6 suggests that the alignment, and thereby the
calibration are impeccable. This is roughly true, but Fig. 6 nevertheless contains a
correction factor. The LDV velocity data had to be multiplied by 0.98, so that the
alignment becomes the best (i.e. the sum of the squares of the discrepancies
becomes minimal). This is shown in Fig. 8.
Does this mean that the LDV method is wrong by 2% (relative to the LIMA),
and the “LDV Scaling factor” should be modified from the factory set 5 mm/s/V to
4.9 mm/s/V? The answer is clearly no. Namely, let’s see if the result of the cor-
rection depends on at which point the laser beam of the LDV hits the backplane of
the corner prism. We have already indicated the measurement points at the back-
plane of the enlarged prism in the upper right part of Fig. 4; we measured the
graphs showed in Fig. 5, 6 and 7 with the laser beam reflected from measurement
point 1 (the centre of the backplane). We summarized the multiplication factors that
we obtained in a similar way for the other measurement points in Table 1.
According to the data in Table 1, moving vertically upwards along the centreline
of the prism the correction factor changes by about 1% for every mm. (The top of
the corner prism swings out more, therefore at the top the LDV data have to be
Comparison of Two Laser Interferometric Methods … 215

reduced more in order to obtain alignment with the LIMA.) Moving horizontally
(also along the centreline) the required corrections are much smaller; the two edges
of the prism vibrate with an amplitude that is about 2% greater than the middle.
This indicates that the applied knock creates a vibration that has a small torsion
component as well.
The laser beam of the LIMA enters and leaves the prism on its other side, at a
distance of 15 mm from each other in its horizontal centre plane. Therefore, the
motion of the reflecting points of the prism could be around the average of the
centre (point 1) and the two edges (points 5–6), for which the corresponding
correction factor could be 0.99. This means that the velocity data measured by the
two devices differ by about 1% if we consider the points of the centreline of the
prism with coordinates (7;0) and (−7;0). At the same time we think that the setting
precision of the spot of the laser beam is approximately 1 mm on the backplane of
the prism, which—as described above—leads to an approximately 1% error for the
correction factor. That is in the end the discrepancy of the data is not larger than our
estimated measurement error. Therefore the scaling factor of the LDV is accurate
within a 1–2% margin of error. This (quite small) measurement error may be caused
mainly by the fact that the LIMA measures the averaged vibrations of the whole
corner prism, while the LDV only measures one point.

4 Conclusions

Based on the measurements carried out we can state that both the LIMA and the
LDV work excellently given the appropriate circumstances and settings. The
modernized data collection system fits the old opto-mechanical unit well.
The new vibration-free table with the air springs significantly improved the
accuracy of the measurements. It was repeatedly demonstrated that laser interfer-
ometric measurements should only be carried out on vibration-free tables of ade-
quate quality otherwise the vibrations of the precision elements caused by the
environment or another element can falsify the measurements.
We found that the LDV is more suited to determine the instantaneous velocities
than the LIMA, which primarily measures displacement. This is especially true for
the vibrations where the velocity remains under 0.1 mm/s. The development of the
measuring program of the LIMA, working out a better method for calculating
instantaneous velocity and acceleration is certainly warranted.
The measurements demonstrated that the different points of the gently knocked
corner prism in the measuring arm vibrate differently. These differences are greater
than the differences of the data provided by the LIMA and the LDV (in the central
setting). Based on our measurements we can state that within the margin of error the
LIMA and the LDV measure the velocities to be the same (with the constraints
mentioned above). We estimate this margin of error to be 1–2%. The reason for
even this (small) error is not within the devices, but it is the fact that the devices
employing completely different principles do not “see” the same exact vibration.
216 M. Béres and B. Paripás

Acknowledgements This work was carried out within the framework of the Centre for
Excellence in Mechatronics and Logistics operating in the strategic research field of the University
of Miskolc.

References

1. Beeck MA, Hentschel W (2000) Laser metrology—a diagnostic tool in automotive

development processes. Opt Lasers Eng 34:101–120. doi:10.1016/S0143-8166(00)00077-4
2. Chen F, McKillip D (2007) Measurement and analysis of vibration and deformation using
laser metrology for automotive application. Proc Institution Mech Eng Part D—J Automobile
Eng 221:725–738. doi:10.1243/0954470JAUTO67
3. Castellini P, Montanini R (2002) Automotive components vibration measurements by
tracking laser doppler vibrometry: advances in signal processing. Measur Sci Technol
13:1266. doi:10.1088/0957-0233/13/8/314
4. Citrovszky A (1998) LI-03 típusú szubmikronos lézerinterferométer kezelési és karbantartási
könyv, Budapest
5. Jani P, Czitrovszky A, Szótér L, Tajnafoi J, Barna B, Molnár L (1999) A laser interferometric
motion analyser system and some of its applications. Publications of the University of
Miskolc, Series D-Natural Sciences Physics 1:17
6. Béres M, Paripás B, Majár J, Bodolai T, Rónai L, Illavszky V (2016) Comparative study of
two laser interferometric methods for the study of vibrations. In: Proceedings 30th MicroCAD
international multidisciplinary science conference, Section C3: natural sciences,
ISBN:978-963-358-113-1, Miskolc (Hungary)
7. (2012) Polytec GmbH: PDV-100 Portable Digital Vibrometer Datasheet
8. Hussain B, Muhammad T, Rehan M, Aman H, Aslam M, Ikram M, Raja MYA (2013) Fast
processing of optical fringe movement in displacement sensors without using an ADC.
Photonic Sensors 3(3):241–245. doi:10.1007/s13320-013-0088-5
9. Ripper GP, Garcia GA, Dias RS, Zindeluk M (2005) Primary calibration of single-point laser
doppler vibrometers (LDV). In: Inter-Noise Congress, Rio de Janeiro
10. Mark Cs (2004) Fundamentals of light sources and lasers. Wiley
Deburring of Polimer and Metal Matrix
Composites

János Líska

Abstract Nowadays have developed more and more such materials, which
mechanical and physical properties are extremely better compared to commonly
used materials. These composite materials, which were used before a few years ago
just in the construction industry for stiffening or as a decorative element, are used
also often is the automotive industry. This fact opened a new way in machining,
where we must comply to increasingly stringent precision and visual requirements.
The burr as attendant phenomenon is known as conventional metal cutting, is also
present at machining of special materials (e.g. composites) as well. For high time
for piece deburring increase the costs of parts production, therefore industry places
much emphasis on reduction of it. The article investigates various options of
deburring after machining of metal and polymer-based composite materials.

1 Overview of Deburring

What constitutes a “burr-free” part varies among companies and quality control
departments. For some, it means having no loose materials at an edge. For others, it
means having nothing visible to the naked eye or an edge condition that will not
cause any functional problem in the next assembly process. Missing material or a
hump of rounded metal at an edge may or may be called a burr [1].
Edge quality is of concern for the performance, safety, cost, and appearance of a
part. The following is a reasonably complete list of the problems caused by
improperly finished edges:

J. Líska (&)
Pallasz Athéné University, Kecskemét, Hungary
e-mail: liska.janos@gamf.kefo.hu

© Springer International Publishing AG 2017 217

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_21
218 J. Líska

• Cut hands in assembly or disassembly.

• Interference fits in assemblies.
• Jammed mechanisms.
• Scratched mating surfaces that allow seals to leak.
• Increased or changed friction.
• Increased wear on moving or stressed parts.
• Electrical short circuits.
• Cut wires from sharp edges and sharp burrs.
• Unacceptable high-voltage breakdown of dielectric.
• Irregular electrical and magnetic fields.
• Clogged filters and ports from loose burr accumulation.
• Cut rubber seals and O-rings.
• Turbulence and nonlaminar flow.
• Reduces formability.
Figure 1 illustrates the most commonly used deburring processes. Unfortunately,
no single machine or process produces all the required edge quality on every edge
for every burr without side effects [1].
Our aim is to decrease the material cost of floor structure by changing the profile,
dimensions and number of cross members as well as the thickness of the deck plate.

Fig. 1 Principal deburring processes and their removal mechanisms [1]

Deburring of Polimer and Metal Matrix Composites 219

2 Deburring Economics

Few examples of comparative deburring economics have been published. The

following equations provide a reason been estimate of costs using various pro-
cesses. The equations use following terminology [1]:
C Deburring cost per part
CD Depreciation cost per hour = Machine cost per operating hours
CM Maintenance cost per hour of operation
CL Labour cost per hour to run the machine
CP Cost of power used
CA Cost of cleaning per hour after deburring
CE Cost of media per hour = media cost
percentage of hourly attrition
CC Cost of compound per hour
CW Cost of water per hour
DO Overhead as percentage of labour rate
N Number of parts run per hour = n/t
n Number of parts run per cycle
t Time (hours) per cycle
Cg Cost of gas per cycle
W Power used, in kilowatts
Ct Total tool cost
Np Total number of parts
Cs Total cost of solution
K1 Percentage of cycle time that an operator actually spends controlling
deburring operation
K2 Percentage of cycle time that operator spends cleaning parts
CB Cost of cleaning materials per hour
For vibratory (loose abrasive) processes, use

C ¼ ½CD þ CM þ WCP þ CB þ CE þ CC þ CW=N þ ½CLð1 þ DOÞðK1 þ K2Þ=N

ð1Þ

For thermal energy methods, use

C ¼ ½CD þ CM þ CLð1 þ DOÞWCP þ CA=N þ Cg=n þ Ct=Np ð2Þ

For brush deburring, use

C ¼ ½CD þ CM þ CLð1 þ DOÞWCP þ CA=N þ Cb=Np1 ð3Þ

220 J. Líska

For flame deburring, use

C ¼ ½CD þ CM þ CLð1 þ DOÞWCP þ CA=N þ Cg=n ð4Þ

For manual deburring, use

C ¼ CLð1 þ DOÞ þ CA=N þ Ct=Np ð5Þ

For mechanical deburring, use

C ¼ ½CD þ CM þ WCP þ CA þ CLð1 þ DOÞ=N þ Ct=Np ð6Þ

For chemical deburring, use

C ¼ ½CD þ CM þ WCP þ CA þ CLð1 þ DOÞ=N þ Cs=Np ð7Þ

For electropolish deburring, use

C ¼ ½CD þ CM þ WCP þ CA þ CLð1 þ DOÞ=N þ Ct=Np þ Cs=Np1 ð8Þ

These equations have some important limitations. First, they assume that the
conventional form of the deburring process is used. As mentioned earlier, it is
frequently possible to alter the process slightly to obtain faster or better results.
Such alterations may insert another cost term into the equation, however. Unless the
conventional approach is used, the equation provides only initial estimates of cost.
A second limitation of these equations is that they assume that the one knows the
value of each component and the time required to remove the burr. Although it is
possible to use “rule of thumb” costs for media, compounds, and the like, only a
few publications provide any information on the time required to remove a burr of a
specific size. As additional research is reported, this will become less of limitation.
In the interim, information can be extrapolated from the results produced by other
parts subjected to the same process.
At third limitation of these equations is that they ignore the costs of floor space,
area heating, lighting, maintenance, insurance, and supervision [1].

3 Burr Formation Mechanism

Burrs, flash, and related protrusions are formed by the six physical processes listed
in Table 1. Burrs formed by the first three processes involve plastic deformation of
the workpiece material. Solidification of material on the working edges, the fourth
processes of formation, forms a burr-like projection. The fifth process, incomplete
cutoff, occurs when the workpiece is allowed to fall from the part before the cut is
completed. Flash forms whenever the pressure on molten material is sufficient to
force the material between the two halves of a die or mould. The examples of burr
Deburring of Polimer and Metal Matrix Composites 221

types we can see on Figs. 2 and 3 shown in the theoretical definition of burr
characteristics [1].
In end-milling operations, Poisson burrs are formed on edge 1. Depending upon
cutter geometry, they also form on edges 2, 4 and 10 (Figs. 4 and 5). An entrance
burr occurs on edge 6. Rollover burrs are produced on edges 3, 7, and 9. On half of

Table 1 The physical processes involved in the formation of burrs, flash, and related protrusions
[1]
Process Name of protrusion
Lateral flow of material Poisson burr
Bending of material (such as chip rollerover) Rollover burr
Tearing of chip from the workpiece Tear burr
Redeposition of material Recast bead
Incomplete cutoff Cutoff projection
Flow of material into cracks Flash

Fig. 2 Examples of burr types; poisson and entrance burr

Fig. 3 Definition of burr

characteristics [1]
222 J. Líska

Fig. 4 Burr locations on

slots produced by side or end
mills

Fig. 5 Identification of burr

locations in end-milling edges
of workpieces [1]

edges 5 and 8 an entrance burr is produced, whereas a rollover burr equals the radial
depth of cut until the cut exceeds 0.6 times the tool diameter. For deeper cuts in the
height tends to remain constant at 0.6 times the tool diameter, although it may vary
with workpiece properties and tool forces [1].
Deburring of Polimer and Metal Matrix Composites 223

Fig. 6 DMG—Sauer
Ultrasonic technology [2]

4 Examples of Composite Machining

The central element of ultrasonic-based machine is specially, for this purpose

developed spindle with tool system, known as an appropriated tool system for HSK
standard. It takes across the high-frequency of ultrasound technology by induction
from the spindle to the Ultrasonic HSK clamping, which puts the tool into vibra-
tional mode. The contact between tool and work-piece creates and terminates fifty
thousand times by oscillation. By using of the technology, the chips become
detached from the surface of work-piece only in small pieces. Figure 6 shows the
conceptual drawing of technology.
The advantage of ultrasonic machining showed at most on work-piece: while
material smeared (negative burr formation) at conventional machining machines,
the surface lost its cell structure. At ultrasonic machining surface preserved its
original appearance.
At designing of experiments is never clean, which results are expected. This case
is a completely good example of that mechanism of stock removal of polymer based
composite materials are very different compared to stock removal of metallic
materials. It is interesting from several aspects, because momentarily do not exists
model, which would accurately describe this process. During the experiments and at
interpreting of results we can conclude, that in that case is no ductul cutting. It is
clearly visible from both images (Figs. 7 and 8). The chip breaks in shear plane and
there is no plastic surface deformation. It can be also observed that two machined
materials behave differently. While at material GFRP shapes of the chip are elon-
gated, shapes of chips at materials CFRP are fragmented. This can be explained by
the fact, that the modulus of elasticity of material GFRP is larger.
224 J. Líska

Fig. 7 GFRP high-speed

camera image

Fig. 8 CFRP high-speed

camera image

References

1. Gillespie LK (1999) Deburring and edge finishing handbook. Society of Manufacturing

Engineers. ISBN 0-87263-501-5
2. Uni-Hannover Teil projekt. http://www.sfb653.uni-hannover.de/en-us/Pages/Teilprojekt-T4.
aspx. Accessed 26 Feb 2016
The Past, Present and Future
of the Training of Internal Combustion
Engines at the Department of Energy
Engineering of BME

Ákos Bereczky

Abstract Vehicle industry plays an important role in the GDP production and
employment of present-day Hungary. In the June of 2016 the share of vehicle
manufacturing subsection within processing industry was 31.4% [1] and the sector
employed 135 thousand persons in 2014 [2]. Thus higher education takes a sig-
nificant role in the provision of adequately trained students and the co-operation
with the industry has also a great role. The author reviews the training of internal
combustion engines, which is the most frequently applied resource today—related
to his own narrower field within vehicle industry and the joint research at the
Department of Energy Engineering of BME. The author presents the Department in
the field, the present areas of research, the industrial co-operations and the available
infrastructure. Finally the further plans are being outlined.

1 Introduction

The target of training at University level is to issue appropriate engineers for

industry professionals and to educate junior lecturers, the next generation of pro-
fessors. The training is based on lectures, problem solving and laboratories. The
output levels are different for a manufacturing or a research and development
company and different for a Ph.D. study. But the relevant theoretical background is
needed to support practices and laboratory measurements. The practices can be a
good complement with modelling practices, while labs are necessary for proper
infrastructure and good industrial support with demonstration, joint projects and
research. The Author presents the present infrastructure and research activities of
the past and present.

Á. Bereczky (&)
Department of Energy Engineering, Budapest University of Technology and Economics,
Bertalan Lajos u. 4–6, Budapest 1111, Hungary
e-mail: bereczky@energia.bme.hu

© Springer International Publishing AG 2017 225

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_22
226 Á. Bereczky

2 Infrastructure

The infrastructure can be separated into different areas in, the Laboratory of the
Department of Energy Engineering. The first important part is the complex engine
test systems. There are three complex systems installed, they are based on
257–300 kW Eddy-current brakes, maximum speed is 8000 RPM. All of these
systems have gravimetric fuel consumption and blow-by measuring systems and
closed cooling systems. One of them is the AUDI Container Engine Brake System.
Two of the systems are for industrial projects, one of them a spark ignition type
engine is installed with an open source ECU for mainly educational purposes and
for student projects. The laboratory has a gas engine based generator set for the
demonstration and test of the energy systems, this is mainly to determine the heat
balance at different conditions and fuels.
We have different small test systems; an Octane Rating Engine, and a CFR
Cetane rating engine (Fig. 1). The Octane Rating Engine is made by the BASF, on
this system performable to the octane number measurement of gasolines according
to a standard necessary for the examinations. We compare the knock intensity of the
fuel to be examined at a constant compression ratio with the similar value of two
reference fuels, beside different air-ratios. The Cetane rating engine used for
measurements is a single cylinder, four-stroke cycle pre-chamber type diesel
engine. The displacement volume is 610 cm3, the speed is 900 RPM which is the
value determined by the ASTM standard, the adherence of which is ensured by an
asynchronous electric engine belt connected to the engine. These systems are very

Fig. 1 Combustion chamber of the cetane rating engine [3]

The Past, Present and Future of the Training … 227

Fig. 2 The portable emission measuring laboratory

useful for educational purposes, for example students can investigate parameters
like excess air ratio influence on the combustion process. The Department has
indication systems to measure the pressure in the combustion chamber too.
The third element of the laboratory is the emission measuring system, we have
one portable emission measuring laboratory (Fig. 2). With this system we can
measure VOC (T.HC.), NO/NO2/NOx, CO, SO2, O2, CO2 and CH4 concentration in
exhaust channels in the field tests. The second system is the HORIBA exhaust gas
analyser, for VOC (T.HC.), NO/NO2/NOx, CO, O2 and CO2 emission.

3 Research Activities

The main targets of the research activities are focusing on renewable energy
resource utilization in compression ignition (Diesel) engines. This work can be
separated to biodiesel fuels investigation and utilization of alcohol fuels.
Alcohol fuels cannot be used directly in Diesel engines, because of their very
low cetane number and due to other problems, therefore they are usually incor-
porated in Diesel oil. The alcohol fuel‘s higher amount of oxygen in a combustion
process reduces emission rates of incomplete combustion products (PM, HC, CO)
and helps to improve the combustion process. The investigated alcohol fuels were
methanol, ethanol and n-butanol. Methanol is a highly toxic alcohol, there are
several methods known for its production:
228 Á. Bereczky

• Carbon monoxide and hydrogen react over a catalyst

• From carbon dioxide with catalytic hydrogenation of CO2 with H2 (Oláh
methanol economy)
• Direct catalytic conversion of methane.
Methanol has a low Cetane number, and it constitutes a problem that it creates an
unstable blend, therefore either emulsifiers are used, or a triple mixture is created
e.g. with biodiesel [4]. There are several methods to produce ethanol:
• Fossil base: hydration of ethylene
• Microbial fermentation of sugars, distillation, and dehydration.
Its utilisation in Diesel engines is problematic similarly to that of methanol, a
stable mixture can be produced only with additives [5]. The n-butanol production of
can be carried out in the following ways:
• commercially produced from fossil fuels,
• Acetone–butanol–ethanol (ABE) fermentation (Clostridium Beijerinckii bacteria).
The advantage of butanol is that it can be incorporated in diesel oil [6]. However,
its production from renewable raw materials requires further developments.
Methanol tested with diesel fuel, rape seed methyl ester and methanol blend up
to 30 V/V% FAME with 10 V/V% methanol ratio [7]. We found high engine load
values of CO concentration reduced by 2–13%, while HC concentrations were
changing in the range of −8 to +18%, through all experimental load points, the NOx
emission increased by 1.5–10% depending on the load and the soot concentration
values were reduced by 13–45% compared to reference Diesel fuel [7]. The
n-Butanol tested with diesel blend up 5, 10 and 20 V/V%. We found that 1500 rpm
at 75% load for B05, B10 and B20 mixtures the CO emissions reduced by 35.7,
57.1 and 71.4%, NOx emission increased in this study by 10.3, 32.3, and 54.4%,
the HC emission concentration increased by 21.4, 71.4, and 214%, and soot
reduction relative to DF at 55.5, 77.8 and 85.1% respectively.
Wet-ethanol investigated in a Dual fuelled engine [8, 9] (Fig. 3). The investi-
gated parameters were premixed ratio (rp), pilot timing and engine load [10]. At
100% load dual-fuel operation with wet ethanol (except for rp = 20%) is beneficial
from the point of view of emission. The hardly after treatable components like NOx
and especially PM emissions radically decrease. The elevated HC emission can be
easily after treated by the means of conventional oxidative catalytic converter. At
higher premixed ratios CO emissions are in the same range, the decrement of NOx
emission is 61% compared to diesel operation. Particulate matter (PM) emission
radically decreases by 95% compared to diesel operation. HC emission increases
significantly and monotonously with increasing premixed ratios.
On the field of biodiesels the laboratory investigated many different methyl
esters and pure oils like: rape seed oil, sunflower oil, waste vegetable oil and very
The Past, Present and Future of the Training … 229

Fig. 3 Work schema of the dual fuelled engine (based on [10])

special non-edible oils, like croton megalocarpus oil methyl ester and jatropha oil
methyl ester of Eastern Africa origin. Croton megalocarpus oil is a less common
biodiesel feedstock [11], it is a tree borne non-edible oilseed of African origin. The
plants are indigenous to East Africa, and are widely found in the mountains of
Tanzania, Kenya and Uganda. Croton megalocarpus seeds contain approximately
32% by weight of oil [12]. Jatropha oil, will become a significant source for
biodiesel production now or in the future. Jatropha is a fast growing plant, which
requires little water or fertilizer; it can survive in infertile soils [13]. Jatropha oil is
mostly found in developing countries, especially in Africa and Asia. Jatropha plants
have a high seed yield which can be continuously produced for 30–40 years. The
oil content in the jatropha seeds is approximately 30–40% by weight [14].
The tests (Fig. 4) with the utilization of Croton oil and Jatropha oil methyl ester
presented that the emissions of CO and NOx were slightly higher for biodiesel
samples than for mineral diesel. THC was found to be reduced at intermediate load
and full load it was similar, the PM emission was lower for biodiesels at all loads
than diesel fuel [11, 12]. In the field of biodiesel, we started the investigation of the
TBK or MOME fuel in a cooperation with the Engine and Vehicle Emission Test
Laboratory, Centre for Environment Protection and Sustainability Research,
Institute for Transport Science Nonprofit Ltd. [15]. TBK is based on rapeseed oil,
trans esterificated with methyl acetate. It is very important to highlight that the TBK
biodiesel is a pure material; it doesn’t have any additive contrary to the two other
standardized fuels, which contain many additives in order to improve their physical
and chemical properties and make them suitable for use in internal combustion
engines. It is also important to mention that the production procedure of TBK is an
invention of three Hungarian engineers (János Thész, Béla Boros, Zoltán Király).
230 Á. Bereczky

Fig. 4 Test system for research activities [7]; (1 VW 1.9 TDI (1Z) Engine, 2 Eddy current
dynamometer, 3 Fuel consumption measuring system, 4 Exhaust gas analyser system, 5 Smoke
meter, 6 Fuel temperature controller, 7 Intake air temperature controller, 8–9 Computers of the
controlling and measuring systems, 10 Piezo transducer, 11 Charge amplifier, 12 Crank angle
speed encoder, 13 High speed data acquisition system, 14 Computers of the indication system)

4 Modelling

Modelling of the working process is very important for educational and develop-
ment purposes. At problems solving we use a one dimensional system, like the
AVL Boost system and the AVL FIRE CFD software package for CFD
calculations.

4.1 The AVL Boost Software System

The Boost software is a program using a one dimensional model, with the help of
which the processes of the given engine can be easily simulated. To run the Boost
software the built up of a 1D model is needed (Fig. 5), and then the adequate
selection of the parameters of the model, the careful setting of the initial and the
boundary conditions is required [16].
The adequate timing of the valve openings and the start-up of the combustion
has a great impact on the results.
The Past, Present and Future of the Training … 231

Fig. 5 The model built up in

the Boost software for dual
fuelled engines [16]

In the case of dual fuel engine we use triple “Vibe” (i = 1, 2, 3) the resultant
combustion law of the fuel in the certain burnt distances can be obtained with the
summary of the previous functions in the proportion of the transmitted heat

Xu ¼ a X1; u þ b X2; u þ ð1  ða þ bÞÞ X3; u ; ð1Þ

where a, b—are constants to be determined, X1, X2, X3—for the chosen burnt
distances, are “Vibe” approximate functions [16]. In the case of a 24 kW load more
peaks can be found in the operation period of greater alcohol proportion (Fig. 6)
thus the two parameter model (dual “Vibe”) is not able to approximate the original
combustion law any more. With the help of the triple “Vibe” the original com-
bustion law is easy to approximate, the model generated with the help of the so
obtained parameters is expectedly to approximate the real processes (Fig. 6).

4.2 CFD Simulations (Based on [18])

For CFD simulation of internal combustion engines we use the AVL FIRE soft-
ware, this is a multi-purpose thermo-fluid CFD software package. AVL’s Engine
232 Á. Bereczky

Fig. 6 Measured and approximated “Vibe” heat release by a load of 24 kW in dual fuel
operation [17]

Fig. 7 The calculated and measured heat release rates in case of biodiesel fuel at different speeds
and loads [18]

Simulation Environment (ESE-Diesel) in the FIRE module is used to define the

geometry and create the meshing for the combustion chamber. To reduce the
computational demand of the simulations, the 3D model is simplified and a less
detailed model is created. The simulations were accomplished with the Extended
Coherent Flame Model (ECFM-3Z) model, which is developed for diesel com-
bustion simulation. This is a combustion model based on a flame surface density
transport equation and a mixing model that can describe inhomogeneous turbulent
premixed and diffusion combustion. The model relies on the ECFM combustion
model, with an improved burnt gas chemistry description compared to ECFM [19].
With this CFD software package the combustion process of different fuels inves-
tigated (Fig. 7).
The Past, Present and Future of the Training … 233

5 Conclusions

The Author presented a triple base of training of internal combustion engines at the
Department of Energy Engineering of BME. On lectures, we should give a strong
background that is based on high level courses of thermodynamics and fluid
dynamics on B.Sc. and M.Sc. level. For problem solving at B.Sc. level students can
work with one dimensional system, like the AVL Boost system and at M.Sc. level
with AVL FIRE CFD calculation. The laboratories are separated, for B.Sc. level the
heat balance test on the gas engine and the investigation of the combustion process
in the Octane rating engine are requirements. On M.Sc. level the complex engine
tests, ECU application are the main targets.
On Ph.D. level a complex renewable fuel test goes on like the presented different
biodiesel and dual fuel engine tests, and may give a basis for the thesis work.
The present infrastructures are not new, but with the help of the University and
the Industrial partners the infrastructures are sustainable and we can further develop
the systems.

Acknowledgements This study was conducted at Budapest University of Technology and

Economics (BME), Hungary and was supported by the Hungarian/South African Intergovernmental
S&T Cooperation Program (Grant: ZA-17/09). The authors are grateful to the AVL GmbH
Company for the possibility to use their simulation software (AVL Boost and FIRE) under the
university partnership program and for the help of the industrial partners, like AUDI Hungaria Motor
Ltd., AVL-Autokut Ltd., Hungarian Suzuki Ltd. and many other companies.

References

1. First release of the Hungarian Central Statistical Office, Industry, June, 2016 (second
estimation). Publication: 12 August 2016
2. Association of the Hungarian Automotive Industry (in Hungarian: Magyar Gépjárműipari
Egyesület, MAGE) Skilled labour is the obstacle of vehicle industry, 10 April 2015
3. Laza T, Bereczky A (2009) Influence of higher alcohols on the combustion pressure of diesel
engine operated with rape seed oil. Acta Mechanica Slovaca 13(3):54–61
4. Lukács K, Bereczky Á (2015) Investigation of utilization potential of different alcohol fuels in
compression-ignition engines. In: Proceedings of the 12th international conference on heat
engines and environmental protection, pp 57–62
5. Matuszewska A, Odziemkowska M (2011) Study on the impact of co-solvent on selected
properties of mixtures of diesel oil with bio-ethanol. CHEMIK 65(6):543–548
6. Siwalea L, Lukács K, Bereczky A, Mbarawa M, Kolesnikov A (2014) Performance,
combustion and emission characteristics of n-butanol additive in methanol–gasoline blend
fired in a naturally-aspirated spark ignition engine. Fuel Process Technol 118:318–326
7. Žaglinskis J, Lukács K, Bereczky Á (2016) Comparison of properties of a compression
ignition engine operating on diesel–biodiesel blend with methanol additive. Fuel
170:245–253
8. Emőd I, Tölgyessi Z, Zöldy M (2006) Alternative vehicle drives: alternative motor fuels—
hybrid cars—fuel-cell drive. Maróti könyvkerekedés és könyvkiadó Kft, Budapest, 232
pp. ISBN:963 9005 738 (in Hungarian)
234 Á. Bereczky

9. István E (2003) The use of fuels as the fuels produced from agricultural products and waste
from internal combustion engines, technical development, environmental studies
(KMFP-00031/2002) (in Hungarian)
10. Lukács K, Bereczky Á (2011) Experimental investigation of dual-fuel diesel engine with wet
ethanol. In: The role of renewable in energy generation: 10th international conference on heat
engines and environmental protection, pp 185–190
11. Kivevele T, Lukács K, Bereczky Á, Mbarawa M (2011) Engine performance, exhaust
emissions and combustion characteristics of a CI engine fuelled with croton megalocarpus
methyl ester with antioxidant. Fuel 90(8):2782–2789
12. Kivevele TT, Mbarawa MM (2010) Comprehensive analysis of fuel properties of biodiesel
from croton megalocarpus oil. Energy Fuels 24:6151–6155
13. Sarin R, Sharma M, Sinharay S, Malhotra RK (2007) Jatropha-palm biodiesel blends: an
optimum mix for Asia. Fuel 86(10):1365–1371
14. Kivevele T, Huan Z, Lukács K, Bereczky Á, Mbarawa M (2016) Impact of antioxidant
additives on the engine performance and exhaust emissions using biodiesel made from
Jatropha Oil of Eastern Africa origin. Res Dev J 32:1–8
15. Szabados Gy, Bereczky Á (2015) Comparison tests of diesel, biodiesel and TBK-biodiesel.
Periodica Polytech-Mech Eng 59(3):120–125
16. Lukács K, Szurdoki Z, Bereczky Á (2013) Dual-fuel engine modeling with AVL Boost,
ENELKO2013 XIV. Nemzetközi Energetika-Elektrotechnika konferencia. Nagyszeben,
Románia, pp 95–100 (in Hungarian)
17. Bereczky Á, Lukács K, Sipos Á (2015) Determine the dual-fuel combustion engine’s law to
the model AVL Boost, OGÉT 2015: XXIII. Nemzetközi Gépészeti Találkozó, pp 32–36 (in
Hungarian)
18. Szabados Gy, Lovas M (2015) Combustion simulation of diesel fuel and biofuel by the help
of AVL’s multi-purpose thermo-fluid CFD software. In: Proceedings of the 12th International
Conference on Heat Engines and Environmental Protection, pp 51–56
19. AVL: AVL FIRE® VERSION (2011) Combustion and emission module. Edition 10/2011
Concept of a New Method for Helical
Surface Machining on Lathe

Dániel Kiss and Tibor Csáki

Abstract In this paper the concept of a new turning method for helical surfaces
will be demonstrated. In mechanical engineering several types of surfaces can be
mentioned, which are difficult to machine for example the thread of a ball-nut or
small workpieces with high pitch threads or helical surfaces. The main difficulty
with these surfaces is that they can’t be machined with a tool that generates the
desired surfaces, because we can’t assure the rigidity for these machining condi-
tions. These surfaces are similar to threads in a certain respect. But we cannot use
the exact threading cycle, because in the threading cycle the geometry of the tool
defines the geometry of the surface of the thread. This new method is based on the
threading NC cycle, but in this case the tool geometry won’t be copied to the
surface. The surface geometry will be defined by the start points (and/or start
angles) of the thread, which can be computed by an appropriate computer program.

1 Introduction

In machinery there are several helical surfaces or helical grooves for example
ball-screws and ball-nuts, worm gears, etc. Some of these are difficult to be
machined. These difficulties can be originated to the shape, size and the position of
these surfaces. Taking a ball-nut for example the inner ball groove is complicated to
cut when it has high-lead profile. These helical surfaces are similar to threads but
have a non-standard profile. There are several solutions to this problem, but these
are not universal solutions in the meaning of tooling. It means that each type or size
needs a tool which can only be used for the machining of that workpiece.

D. Kiss (&)
T. Csáki

University of Miskolc, Miskolc, Hungary
e-mail: kiss.daniel@uni-miskolc.hu
T. Csáki
e-mail: csaki.tibor.dr@gmail.com

© Springer International Publishing AG 2017 235

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_23
236 D. Kiss and T. Csáki

Since the Department of Machine Tools at the University of Miskolc has dealing
with machine tool components and design, we started our investigation around
high-lead ball-nuts [1, 2].

2 Thread Cutting

In modern CNC machines there are special subroutines or cycles for a specific
application. One of these applications is the thread cutting which has several cycles
to fulfil the expectations. From these the L97 or CYCLE 97 is the Sinumerik
version of the thread cutting cycle, which has a wide parameter range to make it
possible the programming of different type of threads.
During thread turning we have to use the appropriate tool for a specific thread. It
follows that during thread machining the profile of the tool defines the profile of the
workpiece. There are 3 main types of infeed methods, which are the following
(Fig. 1).
Based on these we face problems when we want to machine large threads or
threads with special profiles, especially at inner thread machining. In this case the
thread milling can be a solution. But in this case we have to use special tools for
each workpiece, because we have to use modified quill angle [4].

3 Test with Profiled Tool

At the beginning, we wanted to simulate the machining of ball-nuts. We changed

the gothic profile to a simple arc to make the tool design easier (Fig. 2). In this case
the tool holder had 32 mm diameter. All the tests were conducted on a DMG CTX
alpha 500 CNC lathe in the workshop of the Department of Machine Tools.
The CNC lathe is equipped with Siemens Sinumerik 840D control.

Fig. 1 Infeed methods from left to right: modified flank infeed, radial infeed, incremental infeed [3]
Concept of a New Method for Helical Surface Machining on Lathe 237

Fig. 2 The tool used for the experiment [5]

Fig. 3 Marks of vibrations on the surface machined with the profiled tool

During the test we used the Cycle 97 subroutine, which used for the program-
ming of thread cutting. Radial infeed and decreasing ap was programmed
pass-by-pass. The workpiece had 53 mm inner diameter and the pitch was 25 mm.
During machining excessive vibrations turned up (Fig. 3), which lead to inad-
equate surface finish and serrated chips (Fig. 4). It is clear that this kind of man-
ufacturing cannot be realizable at special kind of thread shapes [5].

4 The Concept

Drawing the conclusions based on the previous test, it is clear that there are cir-
cumstances when the conventional threading cycle cannot be used. The length of
the tool edge was too long and the half circle form is far from the linear. Vibrations
appear because of either the rigidity of the machine, workpiece and the tool.
A workable option is to use a tool, which doesn’t have the thread profile on it but
238 D. Kiss and T. Csáki

Fig. 4 The chips in the different stages of the machining [5]

has an elementary geometry. In this case the thread cutting cycle can’t be used in
the conventional way since we don’t use profiled tool. To overcome the problem
that the tool doesn’t machine the full profile we have to examine the threading cycle
parameters. The parameters in the CYCLE97 are the following [6]:
CYCLE97 (PIT, MPIT, SPL, FPL, DM1, DM2, APP, ROP, TDEP, FAL,
IANG, NSP, NRC, NID, VARI, NUMT, _VRT):
• PIT (Thread Pitch)
• MPIT (Thread Pitch as Thread Size)
• SPL (Thread Starting Point Longitudinal)
• FPL (Thread End Point Longitudinal)
• DM1 (Thread Start Diameter)
• DM2 (Thread End Diameter)
• APP (Run-In Path)
• ROP (Run-Out Path)
• TDEP (Thread Depth)
• FAL (Finishing Allowance)
• IANG (Infeed Angle)
• NSP (Starting Point Offset)
• NRC (Number of Roughing Cuts)
• NID (Number of Idle Cuts)
• VARI (Machining Type)
• NUMT (Number of Thread Turns)
• _VRT (Retraction Distance)
Concept of a New Method for Helical Surface Machining on Lathe 239

Fig. 5 Start point parameters

Among these parameters by changing of the TDEP and NSP parameters we can
approximate the desired thread profile.
To calculate these parameters at first the thread profile has to be approximated
with the profile of the tool in the threads normal section. The obtained tool center
points, then have to be transformed to the plane, which is normal to the workpiece
and positioned at the thread start (Fig. 5).
Then the two parameters can be calculated (parameters from Cycle 97):

TDEPi ¼ DM1  ri ; ð1Þ

NSPi ¼ ai : ð2Þ

5 Testing the Concept

To verify the theory we performed a test at the Department of Machine Tools at the
University of Miskolc. In the CNC program we called the threading cycle multiple
times to achieve a simple half circle profile on a shaft. The diameter of the
workpiece was 45 mm, the pitch was 25 mm and the radius of the profile was
4 mm. For the test we used a V shaped tool. The machined workpiece is shown in
the next picture (Fig. 6).
The test was successful, because the machined profile approximated well the
theoretical profile, but there’s much room for improvement in the aspect of surface
finish.
240 D. Kiss and T. Csáki

Fig. 6 Test workpiece

6 Conclusion

In this research, we searched for a possible way to machine threaded surfaces with
non-standard and deep profiles. The problem and a possible solution were shown in
this paper.
The new solution can be used to machine both outer and inner surfaces with the
appropriate tool. This solution applies to the CYCLE97 threading cycle, but can be
interpreted to other cycles where starting point offset can be programmed.
For the further research it is necessary to create a mathematical algorithm which
calculates the required parameters for the cycle and generates the CNC program by
taking into account both the tool and thread profiles. With this program both
roughing and finishing can be executed.
When this algorithm is done a proper tool or tool holder have to be made where
the correct pitch angle can be set.
This new method can be used for the machining of several threaded workpieces
like ball-nuts and -screws, worm in worm drives and automotive steering parts.

Acknowledgements This research was carried out as part of the TÁMOP-4.2.1.

B-10/2/KONV-2010-0001 project with support by the European Union, co-financed by the
European Social Fund, in the framework of the Centre of Excellence of Mechatronics and
Logistics at the University of Miskolc.

References

1. Hegedűs G, Takács G (2013) Tool profile generation by Boolean operations on ball nuts. Key
Eng Mater 581:462–465. doi:10.4028/www.scientific.net/KEM.581.462
2. Hegedűs G (2016) Newton’s method based collision avoidance in a CAD environment on ball
nut grinding. Int J Adv Manuf Technol 84:1219–1228. doi:10.1007/s00170-015-7796-5
3. Sandvik Threading Application Guide (C2920-031) (2015)
Concept of a New Method for Helical Surface Machining on Lathe 241

4. Harada H, Kagiwada T (2004) Grinding of high lead and gothic-arc profile ball-nuts with free
quill-inclination. Precis Eng 28(2):143–151
5. Mihályi G, Csáki T, Kiss D (2013) Design of variations of high-pitch ball nut tool suitable for
machining, XXI. In: OGÉT conference, Arad (Romania), pp 282–284 (in Hungarian)
6. Siemens Sinumerik 840D Cycles Programming Manual (6FC5398-3BP20-1BAO, 01/2008)
 Part III
Electrotechnics, Informatics
Intelligent Transportation Systems
to Support Production Logistics

Péter Veres, Tamás Bányai and Béla Illés

Abstract Intelligent transportation systems (ITS) include both the traffic stream
control and the intelligent vehicles. Cell phone networks and global positioning
systems (GPS) enable the use of geographical information (GI) so that individual
vehicles can locate themselves and global transportation systems can be enhanced
taking advantages of new information technology solutions and algorithms. One of
the major parts of ITS research is the assignment, routing and scheduling of
vehicles in global transportation processes. This paper proposes an integrated
engineering optimization algorithm to support the solution of assignment and
scheduling problems of vehicles in intelligent transportation systems. This novel
approach combines the available hardware and software components of an ITS with
an algorithm to optimize the transportation processes of a global supply chain. To
gain insight into the complexity of the logistic problem, the new model of supply
chain including ITS is also described.

1 Introduction

Successful industrial and service processes provide a significant competitive

advantage against other members of the market. Logistics related operations such as
transportation, warehousing, customer service and inventory carrying account for
10–95% of the total cost, depending on the corporate sector, therefore managers
traditionally think; logistics is a mandatory cost bucket. For that very reason it is
very important to take the logistics related operations and processes under control
and turn logistics into the source of competitive edges. Logistics processes can be

P. Veres (&)
T. Bányai
B. Illés

University of Miskolc, Miskolc, Hungary
e-mail: altveres@uni-miskolc.hu
T. Bányai
e-mail: alttamas@uni-miskolc.hu
B. Illés
e-mail: altilles@uni-miskolc.hu

© Springer International Publishing AG 2017 245

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_24
246 P. Veres et al.

divided into four important sequences: purchasing, production or service, distri-

bution or sales and recycling. Transportation processes are especially remarkable in
the purchasing and distribution processes of supply chain.
The explosion in technologies of ITSs made it possible to build geographical
information systems. The improvement of wireless communication, computational
technologies, floating cellular data, global positioning systems, sensing technolo-
gies and onboard systems leaded to the possibility to locate individual vehicles and
communicate with them real time. This real time communication involves the
collection and analysis of big data and makes it possible to optimize their operation
from the point of view logistics, emission and costs.
This paper is organized as follows. Section 2 presents a literature review, which
systematically summarizes the research background of supply chain optimization,
ITSs and heuristic optimization. Section 3 describes the problem and hypothesis.
Section 4 presents the general model of a supply chain including ITS based vehicles
for transportation operations. Section 5 explains the solution algorithm.
Conclusions and future research directions are discussed in Sect. 6.

2 Literature Review

This section reviews relevant literature related to supply chain optimization, ITS
and heuristic optimization. Due to the large amount of researches on these fields the
most relevant scientific results have to be summarized before to elaborate the
model, algorithm and solution.

2.1 Research on Supply Chain Optimization

Within the frame of this chapter the author focuses on the supply chain optimisation
of service sector. There are two main types of supply chains of service processes. In
the case of service only supply chains (SOSC) the products of services are only
pure services and no real physical products take part in the supply chain operations.
In the case of product service supply chain (PSSC) both services and related
physical products are integrated parts of the processes [1]. The research field of
service related supply chains includes a wide area of services: energy supply chain
[2], supply chain of chemical corporate sector [3], garment industry [4], trans-
portation [5] and local food supply chain [6]. The optimization of supply chain
includes the following main topics: scheduling of logistics services in service
supply chain [7], routing and service level consideration [8], cost optimization [9],
location, inventory and pricing [10]. The researches in the field of routing of supply
chain cover both homogeneous and heterogeneous vehicle routing [11]. The
widening of heterogeneous supply chain caused a big change of design, control and
scheduling of vehicles. This change influences the required transportation and IT
Intelligent Transportation Systems to Support … 247

infrastructure. To the author’s knowledge there has been no research work on

real-time routing and fleet size optimization of supply chain, which takes into
consideration the geographical information from ITS.

2.2 Research on Intelligent Transportation Systems

Intelligent transportation systems represent a group of tools and technologies, by

the aid of which it is possible to improve the management, control and controlling
processes of different transportation systems. ITS technology includes hardware and
software components. Operation research has a key role to reach the full potential of
ITS and there are publications in the field of transportation engineering focusing on
the assessment and contribution of operation research [12–14]. GIS technology
makes it possible to support the optimization of supply chain and define strategies
to minimize the overall costs through the reduction of fossil fuel consumption [15].
ITS include Advanced Traveller Information System (ATIS) and Advanced Driver
Assistance System (ADAS) [16] and these systems can support the design and
control of transportation supply. Wireless technologies, state of the art electronics
and automation makes it possible to build a strong hardware background for
methods and algorithms, by the aid of which it is possible to enhance the possi-
bilities of ITS. ITS can be represented not only in the field of freight transport but
also in the processes of passenger transport.

2.3 Research on Heuristic Optimization

In computer sciences, mathematical optimization and engineering heuristic is a

technique for problem solving and decision making in the field of NP-hard
problems.
The heuristic optimization of transportation supply covers not only the func-
tional parts of logistics (purchasing, production, distribution), but also focuses on
the development of new models and methods to decrease the economical footprint
of logistics services [17]. Heuristic optimization methods can support the solution
of problems in the field of internal logistic related problems and external logistic
networks. The complexity of supply chains required up to date methods to find
near-optimal network design solutions for design, operation and management. The
optimization methods of global supply chain are mostly heuristic methods. Much
literature has been published in the field of design of large scale supply chain. There
are several applications for heuristic algorithms from the sizing optimization of
truss structures [18] to the transport energy modelling [19] or nuclear energy
production [20]. One of the main streams of literature in the field of hybrid
metaheuristic addressed the development of complex algorithms for large systems
[21–23].
248 P. Veres et al.

3 Problem Description

There are many companies, who launch a vehicle that is partially or completely
empty, such as an empty return. It has always been a huge problem for companies,
especially those who have small vehicle fleet and there is no proper transportation
background behind them. In some cases, carrier companies have methods, that can
chance the route of a transport vehicle after launch (for example, a traffic problem
occurs), but this usually requires human intervention and planning. There are
solutions in the form of network-based applications that collects data from joined
small and medium companies and about their vehicles. However, most of these
applications only collect the basic parameters of the vehicles and when they are not
in use so the application is trying to provide work for that vehicle for the specified
period. If we could integrate algorithms and extensions into these systems, it would
track and calculate automatically with vehicles on the road and give them per-
formable tasks, with which we can achieve further savings. This can be imple-
mented by an expanded set of data, which contains the vehicle’s position at every
new transporting task, which requires an integrated GPS and identification system.
It also needs to provide the vehicle’s free capacity, destination, and the delivery
deadline to avoid delays. With such a system it makes easy for a company to
determinate, to give the new task to a running vehicle (Fig. 1) or to launch a new
one to fulfil the demands (Fig. 2).

Current posiƟon (CP) Final desƟnaƟons (FD)

1 1
2 2
3 3
i i
n n

Sub-pickup Sub-desƟnaƟons
Points (SP) (SD)

1 1
2 2

Transport vehicle j j

parking site (PS) m m

1
k
o

Fig. 1 Model of the supply chain related assignment problem without runtime optimization
Intelligent Transportation Systems to Support … 249

Current position (CP) Final destinations (FD)

1 1
2 2
3 3
i i
n n

Sub-pickup Sub-destinations
Points (SP) (SD)

1 1
2 2
j j
Transport vehicle m m
parking site (PS)

1
k
o

Fig. 2 Model of the supply chain related assignment problem with runtime optimization of free
transportation resources

4 Model of Its Based Supply Chain

The model of the geoinformation based supply chain coordination can be divided
into four main parts. The first part is the cooperative level. This level includes all
networking partners for production and services processes. The suppliers are
responsible for the production of components of final products produced by the
main company. Trading companies mean business today to business partners. In the
case of 3rd party logistic partners, the production company outsources logistic
processes, like transportation, loading, warehousing (included bounded warehous-
ing) and forwarding. Other services, like banking, insurance and customs belong to
this level of the model too. The corporation level includes the design and operation
of own resources. The product design and prototyping is the first important part of
this level because this is the initial part of the lifetime of the product, where its
logistic related requirements for production can be influenced. Design processes
(CAD) are connected to process planning (CAPP) and production planning through
material requirement planning (MRP I) and manufacturing resource planning (MRP
II). The planning of production processes influences the manufacturing resources,
so CAPP is connected to computer integrated manufacturing (CAM). The pro-
duction planning and scheduling and the design and operation of internal
250 P. Veres et al.

Operation of networking logistic partners

Cooperative level

Suppliers Trading company 3rd party logistics Banking and insurance

Forwarders Customs authorities Bonded warehouses Other services

Operation of logistic on corporation level

Corporation level

Computer aided financial analysis, deployment control

Expense control Competitive costing

Product design & prototyping Planning of production processes

Total
Production planning and scheduling Enterprise resource planning quality
management

Planning of logistic processes

Logistics processes
Operative level

Purchasing Production Distribution

Recycling

Geoinformation
Geoinformatic system

Enterprise resource
Global positioning Navigation Telecommunication
planning

Database,big data & Spatial models and

Remote sensing Maps & attributes
data mining algorithms

Fig. 3 General model of geoinformation based supply chain coordination

production and logistic resources are connected too. The financial analysis
including expense control and competitive costing and the total quality manage-
ment (TQM) effects on the whole corporation process. On the operation level, the
logistics related processes can be divided into four important parts: purchasing,
production, distribution and inverse logistics (Fig. 3).
The geoinformatics integrates the technologies and know-how of global posi-
tioning, navigation, telecommunication and IT solutions, remote sensing of vehicles
Intelligent Transportation Systems to Support … 251

and loads, database and related solutions for big data problems and data mining,
spatial models, algorithms and decision aspects. The geoinformation based supply
chain control optimizes the supply chain processes on the networking level.
The mathematical model takes into account the part of the problem that deals
with vehicles on the road but has no effect on other activities, such as vehicle under
loading. The model is designed to maximize the savings that can be achieved via
inspecting several vehicles on the road by making a detour transport for a new
demand beside its original task. If that’s not possible or it’s too expensive, a new
vehicle should be launched to fulfil the new request. The basis of the model given
by the distances between each point and it’s proportional costs.
As formula (1) shows, savings can be achieved if a vehicle from it’s current
position (CP) goes to a sub-pickup point (SP), where it picks up goods and then
dumps it at a sub-destination point (SD) and continues its original route to the final
destination (FD). If the total shipping costs from the current position to the final
destination and the costs from the transport vehicle parking site (PS) to SD through
SP are significantly lower than the original routing costs than the additional trans-
portation requirement will be performed by vehicles from PS. Along with these, we
have to add the new vehicles launching and order processing cost. Furthermore, if a
request is fulfilled, we can expect a delivery profit from another company. Formula
(3) and (4) determines, how many vehicles do the task needs total. Formula (5) and
(6) stipulates that nor can’t the volume or the weight of the new demands be more
than the free capacity of the vehicle. Furthermore formula (7) describes that the
vehicle with the assigned new request has to arrive at its original destination no later
than the deadline. The notations used in the paper are described in Table 1.
p h
X  i
SPj
CS ¼ CT1  LCPi
FDi þ CT2  LSPj þ LSDj
PSk

l¼1
  ð1Þ
X
q X
mq
 CT1  LCPi
SPj þ LSPj
SDj þ LSDj
FDi þ COs þ CPt ! max:
s¼1 t¼1

i ¼ 1. . .n; j ¼ 1. . .m; k ¼ 1. . .o ð2Þ

0  q  m; npnþm ð3Þ

p ¼ nþq ð4Þ
X
CAPVTi  VTGi  VTDrj ð5Þ
r2hi

X
Ti  WTGi 
CAPW WTDrj ð6Þ
r2hi

SPj SDj
TCi þ TSPj
CPi
þ TSDj þ TFDi  TRi ð7Þ
252 P. Veres et al.

Table 1 Notations used in the paper

CT1 Original transportation cost, which is proportional with the distance
CT2 Transportation cost of a new transport vehicle
COs Cost of the sth order processing and assigning a new vehicle
CPt Profit from fulfilling the tth demand
LCPi
FDi Distance between the current position of the ith vehicle and the ith final
destination
LPSk
SPj
Distance between kth vehicle parking site and the jth sub-pickup point
LSPj
SDj
Distance between the jth sub-pickup point and jth sub-destination
LCPi
SPj Distance between the current position of the ith vehicle and the jth sub-pickup
point
LSDj
FDi
Distance between the jth sub-destination and the ith final destination
n Number of original transports and vehicles that are on road
m Number of incoming transport demands
o Number of transport vehicles parking sites
q Number of new vehicles needed to fulfill the transport demands
p Number of vehicles will be transporting including the new demands
hi Set of elements of j, that the ith vehicle can transport due to capacity
restriction
CAPVTi =CAPW
Ti Volume/weight capacity of the ith vehicle
VTGi =WTGi Volume/weight of the transported goods in the ith vehicle
VTDj =WTDj Volume/weight of the jth transport demand
TBA Time needed to go from A to B
TCi Current time of the ith vehicle at the CP point
TRi Requested arrival time of the ith vehicle to it’s original destination

5 Heuristic Algorithm for Real-Time Optimization

of Its Based Supply Chain

The above mentioned model and a genetic algorithm based heuristic make it pos-
sible to analyze different transportation problems represented by networking pro-
duction companies. Table 2 describes the data set of the transportation demands.
The geographical information of the additional transportation demands is described
by the GPS coordinates of the sub-pickup points (SP) and the sub-destinations
(SD). These GPS coordinates were generated randomly in the eastern part of
Hungary. The required transportation demand is given in loading units (LU) and the
estimated income is on this value.
The geographical information of the available transportation resources is described
by the GPS coordinates of the free vehicles as current positions (CP). The data set
includes the available loading capacities for each vehicle in loading units (Table 3).
The solution method of the above mentioned multidimensional optimisation
problem is based on Rechenberg’s genetic algorithm and includes the following
Intelligent Transportation Systems to Support … 253

Table 2 Data set for the transportation demands

Nr. GPS coordinates of GPS coordinates of Required Estimated
the SPs the SDs capacity in LU income in €
Latitude Longitude Latitude Longitude
1 48.103 20.142 47.722 20.238 5 1000
2 48.943 20.005 48.459 20.315 4 2000
3 48.356 20.629 48.707 21.049 7 1800
4 48.676 21.403 48.829 21.758 8 1500
5 47.723 21.998 47.908 22.022 2 1200

Table 3 Data set for the available transportation resources

Nr. GPS coordinates of the CPs Available loading capacity (loading unit)
Latitude Longitude
1 45.706 20.165 3
2 46.128 20.009 8
3 45.264 21.548 3
4 45.701 21.684 2
5 46.444 20.153 9
6 48.255 20.645 8
7 46.133 22.369 7
8 46.562 20.098 8
9 48.885 20.061 6

steps: (1) generate assignments between free transportation vehicles and additional
transportation demands as permutation vectors; (2) initiate the first population of the
genetic algorithm; (3) perform the genetic algorithm using reproduction, crossover
and mutation operators until termination condition is true.
The operators were created to balance between exploration and exploitation. The
reproduction operator gives preference to the solutions with the best fitness,
allowing them to pass to the next generation.
The crossover operator produces child solutions from more than one parent
solutions. The mutation is responsible for genetic diversity and it supports to avoid
the convergence to a local minimum or maximum. The crossover operator is a
one-point crossover operator, because the use of multi-point crossover operators is
very complicated in the case of permutation vectors and special constraints. Figure 4
shows the convergence of the algorithm in the case of the mentioned data set, where
the ratio of the reproduction, crossover and mutation operators is 50:46:4.
The results of the optimisation leaded to increased capacity utilisation of the
transportation resources, because the additional transportation demands mustn’t be
performed by an expensive third party logistics provider, who is represented in the
model by the transport vehicle parking site (PS).
254 P. Veres et al.

45000,0
43000,0
41000,0
Average revenue

39000,0
37000,0
35000,0
33000,0
31000,0
29000,0
27000,0
25000,0

101
105
109
113
117
121
1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97
IteraƟon

Fig. 4 Convergence of the algorithm

Table 4 Utilization of the free transportation resources

Nr. Loading Available free loading Additional loading Utilisation
capacity capacity (loading unit) (percent)
1 20 3 – 85
2 25 8 – 68
3 20 3 – 85
4 30 2 – 93
5 25 9 – 64
6 15 8 7 100
7 20 7 – 65
8 20 8 – 60
9 25 6 4 40

As Table 4 and Fig. 5 shows not all of the free transportation resources are used
to perform additional transportation processes because of the limited available
loading capacity, the long distance among free transportation resources and
sub-pickup point or among sub-destinations and final destinations, the value of the
expected revenue and the short distance among transport vehicle parking site and
sub-pickup points.
As Fig. 5 shows, in the case of the given dataset only two additional trans-
portation demands are fulfilled by the available free transportation resources, the
remaining seven are performed by the third party logistics provider.
This genetic algorithm based heuristic real-time optimization makes it possible
to increase the efficiency of the whole supply chain including networking pro-
duction companies with more sites and own transportation capacities.
Intelligent Transportation Systems to Support … 255

Transport vehicle parking 1 2 3

site (PS)

Current position (CP) 1 2 3 4 5 6 7 8 9

Sub-pickup Points (SP) 1 2 3 4 5

Sub-destinations (SD) 1 2 3 4 5

Final destinations (FD) 1 2 3 4 5 6 7 8 9

Fig. 5 One possible solution of the problems represented by the above data set

6 Conclusion and Further Research Directions

This study developed a methodological approach for coordination of supply chain

of production companies. It proposed a model for companies performing their
purchasing related transportations by own resources. Two models were proposed:
the first general model shows the levels of supply chain coordination and the second
model is a mathematical model which focuses on the optimization of the assign-
ment problems. The algorithm is tested with different data sets. The results of this
study can be used to improve the utilization of logistics resources and increase
business efficiency. Within the frame of this paper the multiple transportation
demands are ignored. This representation is near to the reality, but the analysis of
the multiple transportation demands would be useful and this is the direction of our
future researches.

Acknowledgements The described study was carried out as part of the EFOP-3.6.1-16-00011
“Younger and Renewing University—Innovative Knowledge City—institutional development of
the University of Miskolc aiming at intelligent specialisation” project implemented in the
framework of the Szechenyi 2020 program. The realization of this project is supported by the
European Union, co-financed by the European Social Fund.

References

1. Wang Y, Wallace SW, Shen B, Choi T-M (2015) Service supply chain management: a review
of operational models. Eur J Oper Res 247(3):685–698
2. Papapostolou C, Kondili EM, Kaldellis JK (2014) Energy supply chain optimisation: special
considerations for the solution of the energy planning problem. Comput Aided Chem Eng
33:1525–1530
3. Sha M, Srinivasan R (2016) Fleet sizing in chemical supply chains using agent-based
simulation. Comput Chem Eng 84(4):180–198
256 P. Veres et al.

4. Chowdhury MMH, Quaddus MA (2015) A multiple objective optimization based QFD

approach for efficient resilient strategies to mitigate supply chain vulnerabilities: the case of
garment industry of Bangladesh. Omega 57(A):5–21
5. Véronneau S, Roy J, Beaulieu M (2015) Cruise ship suppliers: a field study of the supplier
relationship characteristics in a service supply chain. Tour Manag Perspect 16:76–84
6. Martikainen A, Niemi P, Pekkanen P (2014) Developing a service offering for a logistical
service provider—case of local food supply chain. Int J Prod Econ 157:318–326
7. Liu W, Wang Q, Mao Q, Wang S, Zhu D (2015) A scheduling model of logistics service
supply chain based on the mass customization service and uncertainty of FLSP’s operation
time. Transp Res Part E: Logistics Transp Rev 83:189–215
8. Alaei S, Setak M (2015) Multi objective coordination of a supply chain with routing and
service level consideration. Int J Prod Econ 167:271–281
9. Mvundura M, Lorenson K, Chweya A, Kigadye R, Bartholomew K, Makame M, Lennon TP,
Mwangi S, Kirika L, Kamau P, Otieno A, Murunga P, Omurwa T, Dafrossa L, Kristensen D
(2015) Estimating the costs of the vaccine supply chain and service delivery for selected
districts in Kenya and Tanzania. Vaccine 33(23):2697–2703
10. Kaya O, Urek B (2016) A mixed integer nonlinear programming model and heuristic
solutions for location, inventory and pricing decisions in a closed loop supply chain. Comput
Oper Res 65:93–103
11. Koç C, Bektaş T, Jabali O, Laporte G (2016) Thirty years of heterogeneous vehicle routing.
Eur J Oper Res 249(1):1–21
12. Crainic TG, Gendreau M, Potvin J-Y (2009) Intelligent freight-transportation systems:
assessment and the contribution of operations research. Transp Res Part C: Emerg Technol
17(6):541–557
13. Ko YD, Jang YJ, Lee MS (2015) The optimal economic design of the wireless powered
intelligent transportation system using genetic algorithm considering nonlinear cost function.
Comput Ind Eng 86:67–79
14. Shah N, Kumar S, Bastani F, Yen I-L (2012) Optimization models for assessing the peak
capacity utilization of intelligent transportation systems. Eur J Oper Res 216(1):239–251
15. Zhang F, Johnson D, Johnson M, Watkins D, Froese R, Wang J (2016) Decision support
system integrating GIS with simulation and optimisation for a biofuel supply chain. Renew
Energy 85:740–748
16. Cantarella GE (2013) Day-to-day dynamic models for Intelligent Transportation Systems
design and appraisal. Transp Res Part C: Emerg Technol 29:117–130
17. Hoff A, Andersson H, Christiansen M, Hasle G, Lokketangen A (2010) Industrial aspects and
literature survey: fleet composition and routing. Comput Oper Res 37:2041–2061
18. Degertekin SO (2012) Improved harmony search algorithms for sizing optimization of truss
structures. Comput Struct 92–93:229–241
19. Ceylan H, Ceylan H, Haldenbilen S, Baskan O (2008) Transport energy modeling with
meta-heuristic harmony search algorithm, an application to Turkey. Energy Policy 36(7):
2527–2535
20. Aghaie M, Nazari T, Zolfaghari A, Minuchehr A, Shirani A (2013) Investigation of PWR core
optimization using harmony search algorithms. Ann Nucl Energy 57:1–15
21. Wu B, Qian C, Ni W, Fan S (2012) Hybrid harmony search and artificial bee colony
algorithm for global optimization problems. Comput Math Appl 64(8):2621–2634
22. Fesanghary M, Mahdavi M, Minary-Jolandan M, Alizadeh Y (2008) Hybridizing harmony
search algorithm with sequential quadratic programming for engineering optimization.
Comput Methods Appl Mech Eng 197(33–40):3080–3091
23. Amini F, Ghaderi P (2013) Hybridization of harmony search and ant colony optimization for
optimal locating of structural dampers. Appl Soft Comput 13(5):2272–2280
A New Scheduling Software
for Supporting Automotive Component
Manufacturing

Mónika Kulcsárné Forrai and Gyula Kulcsár

Abstract This paper shows the modeling and solving of a special production fine
scheduling problem of the automotive industry. A new scheduling software has
been developed to create execution plans to the production activities that satisfy the
customers’ demands. The main characteristic of the scheduling problem is that
different types of shared resources (e.g. production lines, attachment points, mold
carriers, tools) have to be allocated simultaneously, and multi-processing tasks have
to be scheduled to meet production orders within strict time limits. To solve the
problem we consider not only the primary technological processes but also the
tool-preparation processes. This inbuilt sub-problem is converted to a special
resource environment by using a problem space transformation procedure. To
minimize the tardiness, we schedule the jobs in the given resource environment. We
elaborated a new solving algorithm that can create the optimal solution for the
sub-problem in polynomial running time. This solution for the sub-problem is
applied to meet the tool-preparation constraints of the full production fine
scheduling problem. An advanced multi-objective searching algorithm solves the
full problem. The paper presents the approach of the developed solving method, the
defined objective functions and the applied neighbouring operators. In the solving
process, all the decision making sub-tasks (assigning, sequencing and timing) are
managed simultaneously. The concrete values of the decision variables are set by a
multi-operator and multi-objective local searching algorithm. The fine scheduling
software can also support inventory control by using special objective functions that
can help to optimize the manufacturing from the point of view of the product type
dependent stock levels. The scheduling problem comes from the plant of Fehrer
Hungaria Járműipari Kft., specialized in vehicle seat products (Mór, Hungary).

M.K. Forrai (&)
G. Kulcsár

Department of Information Engineering,
University of Miskolc, Miskolc, Hungary
e-mail: aitkfm@uni-miskolc.hu
G. Kulcsár
e-mail: iitkgy@uni-miskolc.hu

© Springer International Publishing AG 2017 257

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_25
258 M.K. Forrai and G. Kulcsár

1 Introduction

Today’s globalized market environment compels manufacturers to continuously

improve their competitiveness. The rapid changes in the market conditions and the
increasing customer demands make it necessary to improve the production effi-
ciency and the readiness for delivery. However, in order to achieve cost-effective
production manufacturers strive to maintain low inventories, achieve high utiliza-
tion of resources and reduce the associated costs.
Nowadays, modern enterprise management activities cover all operations of the
production systems. Production engineers and managers utilize ever more computer
integrated application systems to support their decision making. Software systems
for discrete production processes can be classified into four hierarchical groups
according to the different supported fields and time horizons. These application
categories are as follows:
1. Enterprise Resources Planning (ERP)
2. Computer Aided Production Engineering (CAPE)
3. Manufacturing Execution Systems (MES)
4. Manufacturing Automation (MA).
Production planning and control modules of the integrated systems deal with the
allocation of limited resources to production activities in order to satisfy given
customer demands over a well-defined time horizon. Both planning and control
tasks lead to optimization problems, in which the main goal is to create plans that
meet the actual demands and maximize the production performance.
The planning and control of discrete production processes usually take place at
several hierarchical levels. This approach means that the solving processes run in a
hierarchical way by ordering the corresponding decisions according to their relative
importance. Hierarchical production planning uses a suitable optimization model at
each level of the hierarchy. Each model extends the constraints of the problem at
the lower level.
In practice, models for solving these problems cover a very wide range due to
the varying characteristics of different production systems and their business
environments. However, the results of production planning and scheduling in
general cannot be applied directly to the operational production management
because the created plans are based on aggregated resources and they consider only
the primary processes. In these cases, production fine scheduling can solve the
problem. An advanced fine scheduling software is needed to create short-term
execution plans (fine schedules) that cover every important detail and process.
In this paper, we present an advanced scheduling software with the applied
model and algorithms for supporting the automotive seat component
manufacturing.
A New Scheduling Software for Supporting … 259

2 Problem Description and Requirements

The scheduling problem, which will be outlined in this section, is inspired by a real
case study concerning the production system of Fehrer Hungaria Járműipari Kft., a
firm specialized in vehicle seat components (Mór, Hungary). This company pro-
duces different types of seat elements with variable series simultaneously. The
customers (business partners) have very strict delivery due dates. The number of the
product types is increasing because of market trends. Due to these requirements, it
is important to develop better production plans, make more flexible, fine schedules,
and use advanced software at the shop floor level.
The plant produces seat components for different vehicle-assembly enterprises.
These customer enterprises regularly send their product demands as production
orders that include the product types, the numbers of items and the due dates. These
orders have to be fulfilled within strict time limits by using the available
resource-constrained manufacturing system.
The seat elements are made on circle-shaped production lines (Fig. 1). The
system manages all the production orders as a set. In general, the product types can
be manufactured on more than one production line (PL), and a PL can carry out a
specified number of laps (rounds) in one shift. Each PL has a dedicated number of
attachment points that are called positions. Shape carriers are connected to these
positions. The construction of the carriers can be one or two-sided and they
transport the shapes on the path. The shape means a special tool, which is named
mould in practice. The tools can be placed on the left and/or right side of the carrier.
The permitted configurations of the devices are specified by master rules and
technological plans. These regulations determine the assignments and usage of
production lines, positions, carriers, tools, sides and product groups.

Fig. 1 A simplified scheme of the production lines

260 M.K. Forrai and G. Kulcsár

The production lines are characterized by different constructions (number and

type of positions, manufacturing capabilities, etc.), and the intensities (number of
rounds in one shift) can also differ. Each production line has an independent
timetable that specifies the enabled shifts as a work calendar on the actual time
horizon. The shift is the unit-long time interval of the calendar. In the shifts the
settings of the production lines can be slightly modified by exchanging one or more
configurations attached to the positions. This exchanging activity is the basic
control operation of the manufacturing execution system. A configuration consists
of one carrier and its mounted tools. For manufacturing products many suitable
carriers and tools are accessible as shared resources. When we want to activate a
given configuration in a given position of a given production line, we have to
disconnect the actual configuration from the position and connect another prepared
configuration to the same position.
To activate a prescribed configuration, the workshop has to prepare it in
advance. The assembly and disassembly tasks of the configurations are carried out
by skilled workers. Such an additional process takes considerable time, so the
number of the performable preparatory tasks is constrained in the full system during
one shift. The constraints can vary in time.
In connection with the described production scheduling problem the most
important requirements of shop floor management can be summarized as follows:
• The production orders should be filled in time with no exceptions.
• The manufacturing processes should be realized with the least possible number
of configuration preparations and exchanges.
• The accumulation of excessive stock should be avoided.
• The product-type dependent minimum stock levels should be maintained.
• The utilization of the production lines should be maximized.
The main purpose of our project was to develop fine scheduling software in
order to initiate a detailed execution program for the production lines in order to fill
the production orders and to synchronize the configuration preparation and
exchange activities on the shop floor in the manufacturing system. The defined
problem does not fit into any of the known scheduling problem categories, so we
have identified a new combined problem of shop floor scheduling and inventory
management.

3 Scope of the Research

In the literature, different scheduling models can be found (e.g. [1–3]). Many survey
papers are also focused on modelling and solving production scheduling problems
(e.g. [4–6]). One of the main groups of these models is the parallel resource scheme.
Detailed reviews for this topic are given, for example, in [7–9]. The parallel
A New Scheduling Software for Supporting … 261

resource environment consists of machines that represent the fundamental building

block of the system.
Considering production performance, both the allocation of machines and the
sequence of jobs are of great importance. Many scheduling models are known in the
literature, but most of them use only one objective function. Usually one of the
performance indicators related to due date plays the main role in scheduling models
for supporting Make to Order (MTO) manufacturing. Only a few of the models (e.g.
[10–12]) deal with multi-objective cases, which are very important in practice.
Since the existing models in the operation research topic often disregard the
machine processing abilities, limited availability time frames of the machines,
limited logistical capacities, and shared tools, the improvement and extension of the
models are justified. In order to consider the aforementioned important features of
real scheduling problems we have focused on the simulation based scheduling
approach. This paper presents our approach, in which a practice-oriented simulation
based multi-objective searching method is used to solve the addressed fine
scheduling problem of vehicle seat manufacturing by preserving important details.

4 Practice-Oriented Fine Scheduling Based on Simulation

We have elaborated an integrated approach to solve the examined combined

problem of production scheduling and inventory control as a whole without
decomposition. In our approach (Fig. 2), all the decisions are made simultaneously.
Applying the solution in practice, a special searching algorithm has been developed
based on simulation, overloaded relational operators and multiple neighbouring

Loading the
Input Data

Building Adjusting/ Fine

the Editing the Scheduling
Detailed Actual the
Model Solution Production

Scheduling the
Simulating the
Preparatory
Production
Tasks

Exporting Measuring the

the Output Objective
Data Functions

Fig. 2 Fine scheduling approach

262 M.K. Forrai and G. Kulcsár

operators. The core of the solver explores iteratively the feasible solution space and
creates neighbour candidate solutions by modifying the decision variables. The
created candidate fine schedule declares the configuration of carriers and tools to be
used and the product types to be produced in each position of each production line
in each shift.
When the searching algorithm creates a new candidate fine schedule, it has to
decide whether the candidate solution is executable. This means that the required
preparatory activities have to be scheduled in time by considering the constrained
capacities of the skilled workers. If there is no lateness when the needed preparatory
tasks are finished, then the production fine schedule is executable; otherwise it is
not feasible.
To analyse the candidate schedules, the actual values of the objective functions
are evaluated by a simulation that virtually performs the planned operations and
processes in the modelled replica of the real resource environment with capacity
and technological constraints. In this execution-driven simulation algorithm, the
workpieces are passive objects of the model and they are processed, moved, and
stored by the active model objects such as production lines, material handling
devices, human workers and buffers. The numerical tracking of the product units
provides the time data of the manufacturing processes. Consequently the simulation
can also calculate exactly the stock level values of the product types in the planned
time horizon. This is the part of the approach that encapsulates the dependency of
real-world scheduling and inventory control problems. The successful adaptation of
the approach into practice is highly influenced by the efficiency of the simulation
algorithm.

5 Scheduling the Configuration Preparatory Tasks

The configuration-preparation sub-problem of the production fine scheduling can be

modelled independently as a special parallel machine scheduling problem with
time-varying resource availability constraints. To solve this problem, we have
developed a rule-based algorithm that gives an optimal solution from the point of
view of the maximum lateness.
This scheduling sub-problem is converted to a special resource environment by
using a problem space transformation procedure. This transformed problem has a
time-varying number of virtual parallel resources, independent jobs with unit
processing times, integer release times and due dates. To minimize the tardiness, we
schedule the jobs in the given resource environment.
The basic idea of the method is that the actual number of the available worker
teams is calculated and stored in each shift, than we schedule the releasable task
with the earliest due date. The algorithm starts from the earliest shift and goes on
with the enabled shifts in chronological order. If any free worker team can be found
in an intermediate shift, and there are releasable unscheduled tasks, then the task
with the earliest due date will be scheduled by assigning it to any free worker team
A New Scheduling Software for Supporting … 263

and to the actual shift. This algorithm can create the optimal solution of the
sub-problem in polynomial running time.
The created schedule of the configuration preparations specifies exactly what
preparatory task is carried out in which shift. If there is any late preparatory task in
the created optimal solution, then the inbuilt scheduling problem is unsolvable,
consequently the examined full fine schedule is not feasible. If each preparatory
task can be finished in time without lateness, then the solution is suitable to meet the
preparation requirements of the planned production processes.

6 Multi-objective Fine Scheduling

6.1 Scheduling Engine

The core of the scheduling software uses an advanced searching algorithm variant
that iteratively moves from an actual schedule to a candidate’s schedule in the
neighbourhood until the stop criterion is satisfied. To reach and examine the
unexplored regions of the search space the method creates new neighbours of
the base solution. The software provides default searching parameters, but the user
can calibrate the actual values on the main form of the solver (Fig. 3).
To avoid a local optimum the method puts the taboo schedules that have been
examined in the recent past into a taboo list. The taboo schedules are excluded from
the further neighbourhoods. The taboo list works as short-term memory limited by a

Fig. 3 The main form of the solver

264 M.K. Forrai and G. Kulcsár

parameter. A pre-defined value specifies the number of the stored elements. A new
set of neighbours is generated at random successively by using priority controlled
neighbouring operators. An input parameter specifies the number of the neighbours
in the neighbourhood.

6.2 Neighbouring Operators

To generate new solutions the operators can carry out permitted modifications on
the base solution. The applied neighbouring operators are as follows:
• N1 operator removes the currently used configuration from a randomly chosen
position of a randomly chosen production line in a randomly chosen shift.
• N2 operator chooses a production order randomly and schedules the manufac-
turing of its product on the default production line of the current product. If it is
not possible, then the operator chooses and uses an alternative production line.
• N3 operator chooses a late production order randomly and schedules the man-
ufacturing of its product similarly to operator N2.
• N4 operator chooses a late production order randomly and schedules the man-
ufacturing of its product on a randomly chosen suitable production line of the
current product.
• N5 operator chooses the production order that has the highest priority from the
set of late orders and schedules the manufacturing of its product on a randomly
chosen suitable production line of the current product.
• N6 operator includes three different modification activities. When this operator is
called for running, one of three variants is chosen randomly.
The first variant chooses the production order that can be filled with the least
amount of moulds from the set of late orders. The second variant chooses the
production order with the earliest due date from the set of late orders. The third
variant chooses the production order that can be filled on the least amount of
production lines from the set of late orders.
In all three cases, the operator schedules the manufacturing of the product which
is required by the chosen production order. The job execution is assigned to the
suitable production line which is loaded with the least number of planned
setups.
• N7 operator chooses randomly one of the two following modifications:
In the first case, the operator chooses the production order that has the highest
priority from the set of late orders. If all moulds of its product are reserved, then
the operator frees a randomly chosen mould and uses it to schedule the man-
ufacturing of the product on the suitable production line which is loaded with
the least number of planned setups.
In the second case, the operator chooses randomly a production order from the
set of late orders. If all moulds of its product are reserved, then the operator frees
A New Scheduling Software for Supporting … 265

a randomly chosen mould and uses it to schedule the manufacturing of the

product on a randomly chosen suitable production line.
• N8 operator removes a randomly chosen configuration exchange from the
schedule.
• N9 operator chooses the production order that has the maximal lateness and
schedules the manufacturing of its product on a randomly chosen suitable
production line.
• N10 operator chooses the production order that has the maximal lateness and
schedules the manufacturing of its product on the default production line of the
current product. If this is not possible, then the operator chooses and uses an
alternative production line.
• N11 operator chooses the product that will have the maximal shortage compared
to the required minimal stock level at the end of the time horizon. The operator
schedules the manufacturing on the default production line of the chosen
product. If this is not possible, then the operator chooses and uses an alternative
production line.
• N12 operator chooses a product randomly from the set of products which will
have shortages compared to the required minimal stock level at the end of the
time horizon. The operator schedules the manufacturing of the chosen product
similarly to operator N11.
Scheduling the manufacturing of a given product in a given position of the
production line means that the algorithm firstly chooses a shift in which a new
exchange can be carried out without exceeding the prescribed maximal number of
exchanges. The second step looks for the final shift by considering the calendar and
the shift in which the next exchange will be carried out. After determining the actual
time frame, the neighbouring operators select the suitable free moulds and examine
whether there is any prepared configuration that includes suitable moulds. If the
answer is yes, then one of the deployable configurations is chosen randomly for
scheduling, otherwise the algorithm chooses the most appropriate free mould based
on the priorities associated with the moulds previously. If the actual carrier can
mount more than one mould, then the corresponding set of moulds is selected for
scheduling. In this decision situation, the neighbouring operators take also into
account what kind of products can be manufactured in the same group. The
methods of the neighbouring operators cover the handler of the shared available
carriers and moulds.

6.3 Objective Functions

The performance indicators of the generated solutions are calculated based on an

in-built simulation module. This component serves the actual values of the objec-
tive functions. To express the shop floor management’s goals as the criteria of a
266 M.K. Forrai and G. Kulcsár

multi-objective optimization problem, we use eighteen objective functions to be

minimized. These are as follows:
• f1: The maximum product shortage [pc]
• f2: The sum of product shortages [pc]
• f3: The number of tardy production orders
• f4: The number of set-up activities (changes)
• f5: The maximum number of set-up activities in one shift
• f6: The number of product types with surplus
• f7: The sum of product surpluses [pc]
• f8: The maximum product shortage at the end of the time horizon [pc]
• f9: The sum of product shortages at the end of the time horizon [pc]
• f10: The sum of the priorities of tardy production orders
• f11: The maximum priority of tardy production orders
• f12: The number of product types with tardiness
• f13: The maximum product shortage (compared to zero) [pc]
• f14: The sum of product shortages (compared to zero) [pc]
• f15: The maximum tardiness of production orders [shift]
• f16: The sum of the tardiness of production orders [shift]
• f17: The number of configuration preparations
• f18: The sum of unused capacities of the production lines [Standard Production
Unit].
In the model we apply new key performance indicators (KPIs) concerning the
stock levels as objective functions. The essence of the measure of these indicators is
that we merge the manufactured product quantities and the delivered product
quantities in a stock-time function, so at any time the actual value of the stock level
in the case of each product type can be calculated (Fig. 4). Using these values, we
can simply calculate the shortage or the surplus compared to the lower reference

Fig. 4 Stock-time chart

A New Scheduling Software for Supporting … 267

limit or to the upper reference limit. The tardiness of the production orders can also
be easily calculated based on the stock-time chart (Fig. 5). We measure the dif-
ference between the actual stock level and the required quantity at the time of the
accepted delivery due date. This signed value of the product quantity shows the
shortage (negative) or the surplus (positive). The lateness of a given production
order means the time interval that has elapsed between the prescribed deadline and
the real delivery time. If the lateness is negative or zero, then the given production
order is satisfied in time (the tardiness is zero), otherwise the production order is
tardy and its tardiness is greater than zero. Using this approach based on the
individual value of each production order, the maximum and the sum values of the
total set can also be interpreted similarly to the usage of job data and optimality
criteria in classical scheduling theory.
The importance of the declared goals can vary over time, so the software sup-
ports the user in expressing the actual importance of the objective functions by
adjusting the priorities (Fig. 6).

Fig. 5 The tardiness of the production order

Fig. 6 User interface for

adjusting the priorities
268 M.K. Forrai and G. Kulcsár

6.4 Comparing and Qualifying the Candidate Solutions

For managing effectively the full system of the objective functions, in the devel-
opment of the software we used our own previously elaborated method, which is
described in detail in [13, 14]. This model is based on the calculation of the relative
quality of a given solution by comparing it to another solution, considering multiple
objective functions simultaneously.
The formal description of the relative qualification model is as follows:

fk : S ! R þ [ f0g; 8k 2 f1; 2; . . .; Kg; fk ! min ð1Þ

0; if maxða; bÞ ¼ 0
D : R2 ! R; Dða; bÞ ¼ ð2Þ
maxða;bÞ ;
ba
otherwise

X
K
F : S2 ! R; Fðsx ; sy Þ ¼ ðwk
Dðfk ðsx Þ; fk ðsy ÞÞÞ ð3Þ
k¼1

where
S the set of the feasible solutions;
fk the kth objective function to be minimized;
K the number of objective functions;
sx,sy two given solutions;
wk the priority of the kth objective function;
F(sx,sy) the relative quality of the solution sy compared to the solution sx
.
Using the signed value of the function F(sx, sy) we extend the interpretation of
the relational operators to the solutions sx and sy in the S. The definition of this
operator overloading is the following:

ðsy ?sx Þ :¼ ðFðsx ; sy Þ?0Þ ð4Þ

where the question mark ? indicates any of the relational operators (<,  , >,
, =, 6¼). For example, sy is a better solution than sx (sy < sx is true) if F(sx, sy) is less
than zero.
This relative qualification model can effectively solve the comparison of the
candidate solutions in the searching process, so the software is able to realize
multi-objective optimization by taking into account the actual requirements of the
users.
A New Scheduling Software for Supporting … 269

6.5 Advanced Functionalities of the Scheduling Software

The implemented scheduling software based on this approach can automatically

create a detailed solution that specifies what the manufacturing system should
perform in the planned time horizon.
To increase the flexibility and effectiveness of the production planning,
scheduling and control process, we developed an advanced software module for
supporting user interactions. Graphical interfaces provide useful charts, diagrams,
tables and reports to visualize the results.
The full scheduling process can be scrutinized and checked on the screen. At
runtime the user can modify the control priorities and the parameters of the
searching algorithm (Fig. 7).

Fig. 7 Graphical user interfaces of the solver engine

270 M.K. Forrai and G. Kulcsár

Fig. 8 An editor window for supporting direct user interventions

The user continuously receives full information about the actual status when the
scheduling process is running. For example, the forms show the elapsed time, the
current and the best searching steps, the current values of the objective functions
and their tendencies. The software effectively supports the usage of the human
intelligence. It allows the expert user to optionally control the searching process by
modifying the priorities of the objective functions and the neighbouring operators.
In addition, the application also makes it possible for the user to edit the actual
schedule by using available operating tools manually (Fig. 8). Similarly to the
neighbouring operators, the built-in protection ensures that the usage of the manual
editing tools can lead only to feasible solutions.
The editor module and its outlined feature can also play a key role when the
process engineers want to declare mandatory configurations to be used for testing.
In such cases, they can use the same editing tools to express their exact require-
ments, but the checkbox with the title SET-UP PROTECTION has to be ticked
before clicking on the button to activate the modification (Fig. 8). The automatic
scheduler has to take into account these new constraints. For this purpose, we
equipped the solver engine with blocking techniques that make it possible to rec-
ognize and distinguish the modifiable and the protected configuration exchanges.
The software offers many formats to show the aggregated and detailed infor-
mation of the fine schedule. For example, the fine schedule of the production
A New Scheduling Software for Supporting … 271

Fig. 9 An appeared format of the created fine schedule of the manufacturing processes

Fig. 10 The slack-oriented schedule of the configuration preparatory tasks

processes can be displayed as a list of the configuration exchanges with the cor-
responding data that declare which configurations (carriers and tools) have to be
attached to which position in which shift and what kind of product types have to be
produced (Fig. 9). Each attached configuration is used in the position until it is
replaced by another configuration.
The schedule of the configuration preparatory tasks can be shown in a simple
table that specifies exactly what pre-assembly activity (PAC) has to be carried out in
which shift (Fig. 10). Each row means one PAC task. The numbers in the second
and third columns of the table show the constrained time interval bounded by the
identifiers of the earliest shifts and the latest shifts in which the task can be done.
272 M.K. Forrai and G. Kulcsár

Fig. 11 User interface for exporting the selected reports

The results are in the first column, which gives the identifiers of the chosen shifts,
and the same data can be seen in the fourth column in the calendar item (date)
format.
The results of the fine scheduling process can also be exported as reports into xls
(excel) or csv (comma separated values) file format in order to increase the
portability (Fig. 11). The saved data can be loaded into our software or other
applications.
In summarizing the research and development work, we can say that our pro-
duction fine scheduling model integrates many new functional components and
sub-systems, of which some essential items are as follows:
• Multi-objective fine scheduling of the production processes
• Managing the time-varying availability-constrained manufacturing resources
• Managing the shared accessible resources
• Managing the predefined configurations which have to be used in given posi-
tions and given time intervals for the test manufacturing required individually by
process engineers
• Scheduling the preparatory tasks of the manufacturing
• Optimizing the stock levels within the product-typed dependent limits
• Significant expansion of the scope of the objective functions in order to meet
industrial demands.
These advanced functionalities of the presented software effectively help to
satisfy the requirements of shop floor control systems in practice.
A New Scheduling Software for Supporting … 273

7 Conclusions

This paper is primarily concerned with automotive industrial scheduling problems,

which require advanced scheduling software to assign limited available resources to
the operations and to schedule the preparatory and the manufacturing activities over
time. The problem comes from automotive seat component manufacturing, in
which typically variable series of items are produced in parallel. The developed
models, algorithms and the implemented software are used in practice.
This paper describes the application of a practice-oriented approach for solving
multi-objective scheduling problems. It is based on built-in production simulation
and usage of relational operators for comparing the qualities of solutions in
advanced searching algorithms. The introduced software has been successfully
tested on real problems considering the multiple objectives and constraints origi-
nating from the automotive industry. Our new application can schedule predictively
what the actual manufacturing system should perform in the planned time horizon
to fill the current production orders. Each solution created for the shop floor is
feasible because the model considers all of the hard constraints. The software can
also support the management in managing the production scheduling and the
inventory control tasks in an integrated manner. The obtained results encourage us
to apply our approach to solving other planning, scheduling and control problems
of the industry.

Acknowledgements This research was partially carried out in the framework of the Center of
Excellence of Mechatronics and Logistics at the University of Miskolc. The financial background
of the software development was provided by the Fehrer Hungaria Járműipari Kft.

References

1. Brucker P (2007) Scheduling algorithms, 5th edn. Springer, Berlin

2. Pinedo ML (2009) Planning and scheduling in manufacturing and service, 2nd edn. Springer,
New York
3. Pinedo ML (2008) Scheduling theory, algorithms, and systems, 3rd edn. Springer, New York
4. Allahverdi A, Ng CT, Cheng TCE, Kovalyov MY (2008) A survey of scheduling problems
with setup times or costs. Eur J Oper Res 187:985–1032
5. Aytug H, Lawley MA, Mckay K, Mohan S, Uzsoy R (2005) Executing production schedules
in the face of uncertainties: a review and some future directions. Eur J Oper Res 161:86–110
6. Lei D (2009) Multi-objective production scheduling: a survey. Int J Adv Manuf Technol 43
(9–10):926–938
7. Cheng TCE, Sin CCS (1990) A state-of-the-art review of parallel-machine scheduling
research. Eur J Oper Res 47:271–292
8. Koulamas C (1994) The total tardiness problem: review and extensions. Oper Res
42:1025–1041
9. Mokotoff E (2001) Parallel machine scheduling problems: a survey. Asia-Pac J Oper Res
18:193–242
274 M.K. Forrai and G. Kulcsár

10. Baykasoğlu A, Özbakir L, Dereli T (2002) Multiple dispatching rule based heuristic for
multi-objective scheduling of job shops using tabu search. In: Proceedings of the 5th
international conference on managing innovations in manufacturing, Milwaukee, USA,
pp 396–402
11. Loukil T, Teghem J, Tuyttens D (2005) Solving multi-objective production scheduling
problems using metaheuristics. Eur J Oper Res 161:42–61
12. Sbalzarini LF, Müller S, Koumoutskos P (2000) Multiobjective optimization using
evolutionary algorithms. In: Proceedings of the summer program 2000, Center of
Turbulence Research, pp 63–74
13. Kulcsár Gy, Kulcsárné Forrai M (2009) Solving multi-objective production scheduling
problems using a new approach. Prod Syst Inf Eng 5:81–94
14. Kulcsár Gy, Kulcsárné Forrai M (2013) Detailed production scheduling based on
multi-objective search and simulation. Prod Syst Inf Eng 6:41–56
The Context Between the Shift of Average
Demand and the Safety Stock
of Purchased Parts

János Korponai, Ágota Bányainé Tóth and Béla Illés

Abstract In order to optimize the stockpile management costs, the suppliers of car
manufacturers answer by building up a safety stock of different extents to avoid
uncertainties arising from the fluctuation of demands and to minimize their impact.
During the calculation of the safety stock in the case of both the periodic and the
continuous review models, the relations proceed from the system of conditions,
according to which the average level of forecasted demands does not change with
the progress of time. In practice, however, we can see a certain shift, thus, during
the definition of safety stocks and the order of purchased parts, the historical data
can only be used by knowing the direction and the extent of the shift. During our
analysis, we examine the impact of shifts of different directions and extents on the
stock of purchased parts, and we build up a model that predicts the probability of
the occurrence of a stock shortage of an unplanned extent and of an overstocking.

1 Introduction

Unlike in several industrial sectors, the automobile industry is characterized by the

fact that in the supply chain the demands of suppliers in direct contact with the car
manufacturer can be well planned. This means that car companies provide
long-term forecasts for their suppliers, and the demands can change only to the
agreed extent for a short term. These demands, however, do not reflect completely
the market mechanisms; predictability and plannability are in many cases related to
the capacity of the car manufacturers, thus the demands themselves imply stochastic

J. Korponai (&)
Á.B. Tóth
B. Illés

University of Miskolc, Miskolc, Hungary
e-mail: janos.korponai@gmail.com
Á.B. Tóth
e-mail: altagota@uni-miskolc.hu
B. Illés
e-mail: altilles@uni-miskolc.hu

© Springer International Publishing AG 2017 275

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_26
276 J. Korponai et al.

features. The fluctuation of demands influences the supplier’s production demand

and, implicitly, the stocks of purchased parts [1–3].
The stock level of a company’s purchased parts is the consequence of the joint
action of several factors. It is influenced by the quantity and temporal scalability of
each phase of the supply chain, the features of the continuity of production and
supply, and to a great extent, by the frequency of purchases and the individual order
quantities as well. The objective of the logistics management is to guarantee the
stock level required for the adequate handling of production at the lowest possible
level of costs [4–6].

2 The Analysis of Planning Reliability by Comparing

the Planned and Actual Ordering for the Past Period

The calculation of the safety stock takes into account the fluctuation of previous
demands, the forecasted demands and the stock replenishment time of a length
agreed upon with the suppliers. In practice, however, we can see a certain shift, e.g.
in case the customer demands are under-forecasted or if there is a significant
deviation in the production process from the planned percentage of rejects. The
result of these impacts is that the previous use and the future demands cannot be
interpreted as a single continuous series of data, thus, during the definition of safety
stocks and the order of purchased parts, the historical data can only be used by
knowing the direction and the extent of the shift [7, 8].
By knowing the historical data, we can describe the observable regularities in the
data line and we can forecast the future development of the phenomenon. The
development of the future values of a time series is usually the result of the joint
action of several factors, thus the possible outputs can be defined with random
variables, from which only one value will be realized.
The fact that the demands can be well forecasted is characteristic of the auto-
mobile industry. On the short term, major changes take place only at certain actors
of the supply chain due to unexpected disturbances. However, it is a common
phenomenon that the forecasted demand continuously differs from the actual uti-
lization. This reverberates through the production demand towards the supplier
through the issued order and the forecast. The purchase demand for purchased parts
is influenced not only by the deviation derivable from the customer; it is also
influenced by the deviations observable in the production processes within the
company. These external and internal effects—e.g. the deviation in one direction of
the percentage of production scrapes, or the conservative order forecast received
from the customer—can jointly generate a deviation between the demand forecasted
toward the supplier and the actual order. The deviation can be bidirectional, and
underplanning and overplanning of a certain extent can occur as well.
We can see in practice that the forecasted customer demand in relation to time
can be divided into three successive periods:
The Context Between the Shift of Average Demand … 277

• Short-term scheduling
• Medium-term forecast
• Long-term forecast [9, 10].
In practice, the question could arise as to which periods and which forecast
related to which date to compare when comparing the planned and actual utiliza-
tion, since several forecasts were valid for the same period. Accordingly, the
question is what to consider as being the basis of reference. On the basis of the fact
that the tracing of the eventual change in demands after the placement of the order
can happen with the rescheduling of the planned time of the arrival of goods and the
modification of the ordered quantity, the focus of our analysis is the examination of
the stock replenishment time between the placement of the order and the arrival.
First, we must define the relation between the forecast and the actual utilization,
thus the reliability of the forecast. By drawing up for the given article the demand
forecasted for the stock replenishment time for the different order periods, and by
assigning the actual utilization for the same time, we have a developing pattern
(Fig. 1). By considering a longer period, this pattern can reflect the constant
underplanning and overplanning, and the quasi-stationary and non-stationary
fluctuations. The ad hoc changes of signs can be smoothened out by comparing the
cumulated value lines of the planned and the actual utilization demand, the real
tendencies can be represented, and the rate of under- and overplanned periods can
be quantified. Our analysis looks for the solution guaranteeing a safe planning when
the actual utilization systematically exceeds the forecasted demand.
When the demand is overplanned, or the actual utilization decreases after the
definition of the purchase demand of purchased parts and the placement of the
order, a larger stock than the actual demand is purchased. In case the order quantity
is not reduced and the planned arrival of goods is not rescheduled parallel to the
decrease of demand after the placement of the order, the stock level will grow at the
end of the period, and the production supply is guaranteed. The subject of our

Fig. 1 Demand changes, development of planned and actual demand in relation to time [personal
editing]
278 J. Korponai et al.

Fig. 2 Past actual utilization and the development of future expected demand in relation to time
[personal editing]

analysis is the examination of the cases, when the satisfaction of production could
be in danger, thus the case of overplanned demand is not part of the analysis.
After comparing the planned and actual utilization for the past periods, we shall
proceed with the representation of past and future data lines. Figure 2 shows the
actual utilization from the past and the expected demand for the future; the planned
utilization can be represented as two further sections. In the period closer in time,
the utilization demand can be regarded as fixed and frozen, and the demand after
that can vary to different extents. The length of the frozen period can vary in
practice, however, for the analysis, we separated two versions. In the first case, the
frozen period is longer than or identical with the stock replenishment time, which
implies that we can completely calculate with deterministic values when defining
the required quantity of purchased parts, thus the stock coverage is guaranteed
throughout the whole stockpiling period. In case the frozen period is shorter than
the stock replenishment time, the stockpiling mechanism carries uncertainty. The
cause of this uncertainty is that in a part of the planning period, thus the period after
the frozen period, the demand is probably underplanned. In case there is an
underplanning, the stocks would drop to zero before the end of the stockpiling
period or before the arrival of the ordered quantity as a consequence of a demand
increase to a realistic level.

3 Definition of the Trend Function that Gives the Most

Accurate Estimate with an Alignment Analysis

After presenting several past stockpiling periods and the planned demands,
quasi-stationary and non-stationary fluctuations can occur. In order to define the
extent of the underplanning to calculate with, comparable values must be obtained
from the time series. The average and the standard deviation can be the simplest
The Context Between the Shift of Average Demand … 279

features of the time series, thus the average value and the standard deviation of the
actual utilization of the examined past period as well as the average and standard
deviation of the expected demand can be defined. However, these features do not
carry any information about the trends within these time series. Continuous
increase, decrease and hidden trends can occur during a long period. Accordingly, if
we characterize a time series only with the average and the standard deviation, we
may draw wrong conclusions [11].
For a more exact description, we must define the basic trend of the time series.
The aim of trend calculation is the presentation of the basic trend by evening the
time series, and by eliminating the periodical fluctuations and the accidental factors.
We can even the time series with the method of the moving average, analytic trend
calculation and graphic representation. Since the examined time series becomes
shorter as a result of the moving average method, and the graphic representation for
more articles is time-consuming and provides only approximate information, the
analytic trend calculation has become the focus of our analysis.
As a first step of trend calculation, we must decide the type of the function, we
wish to use to estimate the trend of the examined time series. The regression
function describing the development tendency of the time series the best can be
defined as a result of graphic representation. However, in case of more articles, this
is not a viable option in practice, thus we analyse the basic trend for past data lines
with different functions instead of one function type, and we define the type of the
function with the best estimate by deriving from the result of the goodness of fit
used for the different functions. The most common functions for trend calculation
are the linear, the exponential and the parabolic trend function. We will present the
first two function types.
The equation of the linear trend function can be defined by the following formula
[12–15]:

^ytðkÞ ¼ k0 þ k1
t; ð1Þ

where
^ytðkÞ the trend value of the t-th element in case of a linear trend function,
k0 the value of the basic trend at a t0 time in case of a linear trend function,
k1 the slope of the linear trend function, thus the extent of the average growth for
a time unit per one period,
t the series of periods at equal distances from each other expressing the time
variable.
The following relations can express the standard equations necessary for the
initial value of the basic trend and the slope of the trend function [13]:

X
n X
n
y t ¼ n
k 0 þ k1
t; ð2Þ
t¼1 t¼1
280 J. Korponai et al.

X
n X
n X
n
t
y t ¼ k0
t þ k1
t2 ; ð3Þ
t¼1 t¼1 t¼1

where
yt the value belonging to the t-th period of the analyzed series,
n the number of analyzed periods.
Since from the factors of the system of equations we know the values
y1ðkÞ ; y2ðkÞ ; . . .; ynðkÞ , and the dates t and the value of n, the unknowns k0 and k1 can
be defined by solving the system of equations as a first degree system of equation
with one unknown.
The equation of the exponential trend function can be defined by the following
formula [12–15]:

^ytðeÞ ¼ e0
et1 ð4Þ

where
^ytðeÞ the trend value of the t-th element in case of an exponential trend function,
e0 the value of the basic trend at a t0 time in case of an exponential trend
function,
et1 the average relative change per time unit of the exponential trend function.
In case of an exponential trend function, the normal equations can be expressed
with the following relations [13]:

X
n X
n
log yt ¼ n
log e0 þ log e1
t; ð5Þ
t¼1 t¼1

X
n X
n X
n
t
log yt ¼ log e0
t þ log e1
t2 : ð6Þ
t¼1 t¼1 t¼1

Unknowns e0 and e1 can be defined by solving the system of equations as a first

degree system of equation with two unknowns.
By carrying out the trend calculated separately for the times series of past actual
utilizations and future expected demands, we must define the most accurate
regression function describing the alteration of the demand of the examined article
in relation to time. The demand calculable as previously recorded and due during
the frozen period can be regarded as actual utilization, thus the time series can be
extended until the end of the frozen period during the quantification of the past data
(Fig. 3).
When defining the deviations from the actual trend, the residual standard
deviation and the relative residual standard deviation must be quantified for both
The Context Between the Shift of Average Demand … 281

Fig. 3 Past actual utilization and the development of future expected demand by presenting the
linear and exponential trends in relation to time [personal editing]

regression functions. The residual standard deviation shows the quadratic sum of
deviations from the values according to the trend of the time series values [12–15]:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn
t¼1 ðyt  ^ ytðkÞ Þ2
seðkÞ ¼ ; ð7Þ
n

where
seðkÞ value of residual standard deviation in case of a linear trend function.
The relative residual standard deviation shows the deviation of the value esti-
mated with the defined trend function from the actual value. By comparing indi-
cators quantified by applying several types of functions, we can decide which trend
function is the most accurate for the examined time series [13]:
seðkÞ
VeðkÞ ¼ ; ð8Þ
yt

where
VeðkÞ value of the relative residual standard deviation in case of a linear trend
function,
yt mean value of the examined series.
The value of the residual standard deviation in case of an exponential trend
function can be defined with the following relation [12–15]:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn
t¼1 ðyt  ^ ytðeÞ Þ2
seðeÞ ¼ ; ð9Þ
n
282 J. Korponai et al.

where
seðeÞ value of residual standard deviation in case of an exponential trend function.
The value of the relative residual standard deviation in case of an exponential
trend function can be defined as [13]:
seðeÞ
VeðeÞ ¼ ; ð10Þ
yt

where
VeðeÞ value of the relative residual standard deviation in case of an exponential
trend function.
The function type aligning better to the examined time series is the one, where
the residual standard deviation is smaller, thus where the quadratic mean of the
deviations of the time series values from the trend values shows a more favourable
value. Similarly, the alignment with the relative residual standard deviation can be
examined as well, in case of which the more accurate trend function is the one
showing a lower value [16, 17].

4 Forecasting the Future Expected Demand Based

on the Past Actual Utilization, the Definition
of Safety Stocks

After the function type was selected as a result of the alignment analysis of the time
series and the actual trend, the trend value and the equation system of the selected
function can be given to every single period of the time series. During our analysis,
we aim to define the extent of underplanning, thus we wish to quantify the demand
that is not predicted at the time of the order of purchased parts, but which is
expected to occur in the knowledge of past data. This requires the anticipation of
the trend calculated from the actual utilization demand, thus the definition of the
expected data line that assume the future continuation of the discovered regularities;
and during the extrapolation, we calculate the expectable trend values for the future
periods as well (Fig. 4).
We can outline the extent of the underplanning in the knowledge of the projected
trend values and the trend values calculated from the planned demand. The trend
values resulting from extrapolation can be compared with the actual planned
demand of the different periods, and with the trend values calculated for the planned
demands. Figure 4 shows both versions, Dd shows the comparison of trends, Ddtn þ m
shows the difference of trend values belonging to a tn þ m time period.
The Context Between the Shift of Average Demand … 283

Fig. 4 Forecasting the future expected demand based on the past actual utilization by applying the
most accurately aligning trend function [personal editing]

The formulae can be substituted with concrete demands:

y t ¼ dt and ^yt ¼ d^t ; ð11Þ

where
dt the demand belonging to the t-th period,
d^t the trend value of the demand for the t-th period in case of the chosen trend
function.
We can define the value of the underplanned demand for the s stock replen-
ishment time by considering the length of the frozen period (tf Þ and the extent of the
related deterministic demand (dtf Þ:

Dds ¼ d^s  dtf  dstf ; ð12Þ

where
Dds demand for the stock replenishment time,
d^s the value of the demand for the stock replenishment time forecasted with the
trend function,
dtf the value of the demand for the frozen period,
dstf the planned value of the demand belonging to the stock replenishment time
period exceeding the frozen period.
There are two possible directions to reduce the risk caused by underplanning:
• Ordering an increased quantity in the knowledge of the forecasted trend
• Rescheduling the planned quantity for an earlier arrival date.
Given that the conclusion from a forecast carries a certain extent of uncertainty,
it can happen that the underplanned demand actually remains on a low level,
284 J. Korponai et al.

thus a potentially increased order quantity can lead to a durable increase of the stocks’
level or to the formation of obsolete stocks. Therefore, we do not recommend this
version in practice. However, we must underline its advantage, since the arrival date of
the ordered quantity does not change and the delivery is more plannable.
On the other hand, the risk defined by considering the trend values can be avoided
by rescheduling the arrival date of the planned quantity. In case the forecasted
demand is realized at a lower level, the rescheduling of the arrival date of the ordered
quantity to an earlier date will only cause a temporary increase of the stock level,
thus the stock is available at an earlier date due to the arrival of the originally planned
quantity. The advantage of rescheduling is that the forecasted demand was not
increased, thus the increase of the durable stock level and the formation of obsolete
stocks can be avoided. A disadvantage of the model is that an arrival of goods
scheduled for an earlier date can mean the dispensing of the supply chain on the one
hand, and an auxiliary delivery demand that differs from the scheduled delivery
frequency and the generation of additional charges on the other hand.

5 Conclusions

The risk caused by underplanning can lead to supply problems in the satisfaction of
production. In practice, the covering of the production demand for the stock can be
guaranteed through the following steps:
• The analysis of planning reliability by comparing the planned and actual
ordering for the past period.
• In case of insecure planning, the presentation of the past and planned demands
of the examined article, as well as of the frozen period and the stock replen-
ishment time.
• The definition of trend values related to the time series of past and planned
demand by applying different regression functions.
• Definition of the trend function that gives the most accurate estimate with an
alignment analysis.
• With the selected trend function, forecasting the past time series for a future
period.
• Quantification of the quantity between the planned demand and the demand
forecasted with extrapolation regarding the stock replenishment time, by con-
sidering the utilization demand of the frozen period.
• Rescheduling the order date and the arrival date in order for the stock to be able
to cover the demand to be realized at an increased level and on time.
Beside linear and exponential trend functions, other function types can be used
to analyze the demands. Furthermore, the model can be further refined by con-
sidering seasonal fluctuations and accidental factors, so that the impacts of planning
uncertainties can be further moderated.
The Context Between the Shift of Average Demand … 285

Acknowledgements The described study was carried out as part of the EFOP-3.6.1-16-00011
“Younger and Renewing University – Innovative Knowledge City – institutional development of
the University of Miskolc aiming at intelligent specialisation” project implemented in the
framework of the Szechenyi 2020 program. The realization of this project is supported by the
European Union, co-financed by the European Social Fund.

References

1. Krampe H, Lucke H-J, Schenk M (2012) Basics of logistics (in German). Huss-Verlag
GmbH, München
2. Kummer S, Grün O, Jammernegg W (2009) Basics of purchasing, production, logistics
(in German). Pearson Studium, München
3. Vörös J (2010) Production and service management (in Hungarian). Akadémia Kiadó,
Budapest
4. Chopra S, Meindl P (2007) Supply chain management: strategy, planning, and operation.
Pearson Prentice-Hall Publishers, New York
5. Juxiang W, Lijuan R, Changju C, Guozheng L (2006) Influence of different forecasting modes
on automobile manufacturing supply chain. In: Proceedings of the 2006 IEEE international
conference on service operations, logistics and informatics, SOLI 2006, pp 775–779
6. Kása R, Gubán Á (2015) Business process amelioration methods, techniques, and their
service orientation: a review of literature. In: Vastag G (ed) Research in the decision sciences
for global business: best papers from the 2013 annual conference of the European Decision
Sciences Institute. Pearson Education Limited, New Jersey, pp 219–238
7. Yamazaki T, Shida K, Kanazawa T (2016) An approach to establishing a method for
calculating inventory. Int J Prod Res 54(8):2320–2331
8. van der Veen B (1986) Safety stocks and the order quantity that leads to the minimal stock.
Eur J Oper Res 27(1):34–49
9. Koltai T (2009) Production Management (in Hungarian). Typotex Kiadó, Budapest
10. Stock JR, Lambert DM (2001) Strategic logistics management. McGraw-Hill Higher
Education, Boston
11. Cselényi J, Illés B (2006) Planning and controlling of material flow systems (in Hungarian:
Anyagáramlási rendszerek tervezése és irányítása). Miskolci Egyetemi Kiadó, Miskolc
12. Shao J (2005) Mathematical statistics: exercises and solutions. Springer Science + Business
Media, Inc.
13. Korpás A (ed) (1997) Elementary statistics II (in Hungarian: Általános statisztika II.).
Nemzeti Tankönyvkiadó, Budapest
14. Browder A (1996) Mathematical analysis—an introduction. Springer, Berlin
15. Triola MF (2013) Elementary statistics. Pearson Education
16. Triola MF (2015) Essentials of statistics. Pearson Education Limited, Essex
17. Devore JL (2016) Probability and statistics for engineering and the sciences. Cengage
Learning, Boston
An Overview of Autonomous Intelligent
Vehicle Systems

Daniel Tokody, Imre János Mezei and György Schuster

Abstract Vehicles, whose functions are enriched with attributes to increase safety,
environmental awareness, effectiveness, comfort level and prestige, so that they can
play a key role in creating optimal mobility, are now being invented, planned and
manufactured for general use. Throughout the full spectrum of transport, vehicles
will soon exempt people from the routine of driving. If people do not need to drive
their cars, will their driving skills deteriorate or will they entirely fail to develop this
skill later on? Is this threatening us in the near future? What are the latest research
results and regulations on autonomous vehicles? What are the actual advantages of
vehicle automation? We are trying to find the answers to these questions in our
article, while analysing and systematizing information from the national and
international literature on the development of intelligent vehicles by examining the
interaction between various ground transport vehicles, and the related developments
on the subject. Our goal is to create automatic intelligent vehicle systems, within the
concept of intelligent infrastructures and smart cities. The paper provides an FMEA
analysis of intelligent vehicles. To decrease the explored deficiencies in the present
system, applicable proposals are formulated about development areas, such as
forming a communication between vehicular traffic and railed vehicles. We feel that
such developments are important steps in increasing traffic safety, and we regard
them as elements of intelligent transport.

Keywords Automation Self-driving cars Levels of automation Automatic





train operation Grades of automation Intelligent vehicles Smart sustainable



city Intelligent infrastructure V2X

D. Tokody (&)
I.J. Mezei
G. Schuster

Doctoral School on Safety and Security Sciences, Óbuda University, Budapest, Hungary
e-mail: tokody.daniel@mav.hu
I.J. Mezei
e-mail: mezei.janos.imre@gmail.com
G. Schuster
e-mail: schuster.gyorgy@kvk.uni-obuda.hu

© Springer International Publishing AG 2017 287

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_27
288 D. Tokody et al.

1 Introduction

According to the 2014 traffic accident statistics, the primary causes of road acci-
dents are the drivers’ faults (Drivers’ fault total 14 616). This means that 14 161 out
of 15 847 accidents are caused by the drivers, which makes 92% of all accidents
arise directly from human error. In less than 0.5% of all cases, the primary cause is
the vehicle’s technical problem [1].
The facts are the same regarding the accidents between road vehicles, locomo-
tives and rolling stock. There were 19 collisions of road vehicles and railway
vehicles, suburban trains and trams in 2014 (Collisions of public road vehicles with
rail vehicles) [1].
Based on the data found in the statistics, we can increase traffic safety, if we aim
to develop a system that supports the driver in the transport process. Such systems
could be driver assistant solutions, different awareness and attention maintenance
tools or even a collision avoiding emergency brake system. The development of
such systems is already in progress, but in 2010 the importance of developing an
intelligent transport system was stated by a European Directive.
“Intelligent transport systems are such advanced applications that aim to offer
innovative services to a variety of transport modes and in relation to traffic man-
agement, without the actual incarnation of intelligence, as well as enabling different
users to get better information on how to use the transportation network safer, more
coordinated and in a smart way” [2]. This definition has changed in its essential
elements in the past five years. According to the latest research, in many ways we
need to use artificial intelligence in order to develop fully autonomous cars.
Therefore, it is not entirely about creating a fully integrated co-operative system.
We think that in the future the fully autonomous transport system will create a
living, intelligent technosphere in the whole world.
Implementing cooperation between the various participants in traffic is one of the
main elements of the intelligent system. This was the pioneering concept of the
cooperative intelligent transport system.
In our view, ground transport development does not follow a single direction.
This means that the development of road and rail transport systems do not have
common sections. It is also uncommon to put a well working practice (for example
communication protocols, management systems, etc.) from one transport sector to
the other. Therefore a cooperative collaboration—best case scenario—can only be
realised through an interface that provides simplified connection.
The road–rail level crossing point can be a significant threat to vehicles. Still
researching on the subject does not cover common system development, while it
would be critical regarding the cooperative intelligent transport system’s traffic
safety. There is no better proof for the separation of disciplines than the standards
that only deal with communication between road vehicles and road infrastructure.
They do not include the possibility of communicating with other transport opera-
tors. In a few cases, though, they mention tangentially the importance of developing
the communication between different vehicles. Regarding the potential global goals
An Overview of Autonomous Intelligent Vehicle Systems 289

and long term solutions, they do not really bring any solution in the topic of road
and rail transport. With our article, we aim to show the possibility of achieving a
complex, multi-vehicle operation in the intelligent traffic systems, not forgetting the
two main participants in ground transport, that are the road and rail vehicles.

2 The Situation of Intelligent Vehicles in Europe

Today’s research in the field of robotics is dealing with systems that are capable of
autonomous operation, are equipped with intelligent sensors and are able to adapt to
environmental changes. Autonomous intelligent vehicles are basically robots like
these systems, and while developing them, we need to use the results of many
different disciplines. Robotics is a discipline that includes other disciplines, and it is
the most advanced field of automation. In other words, vehicle automation is being
in progress with great efforts [3].
Today’s modern robotic systems, such as autonomous cars are capable of pro-
gressing in convoy or even achieving an objective on their own. Driverless,
autonomous vehicles can be categorized into three groups: UGV (Unmanned
Ground Vehicles), UMV (Unmanned Marine Vehicles) and UAV (Unmanned
Aerial Vehicles) [3]. In our article we are dealing with the possibilities of creating a
complex transport system consisting of UGV vehicles.

2.1 Autonomous Unmanned Ground Vehicle

In the case of ground-level vehicles, our research involves autonomous automobiles

and fixed-track vehicles, capable of automatic operation.
Among fixed-track vehicles, not only heavy railways, subways, trams, suburban
railways and trolley buses should be considered.
In Sweden, Scania has just recently introduced (22nd June 2016) their first true
parallel hybrid trucks (using electricity and biofuel), which can be powered on the
motorway by an overhead line [4]. Naturally, development has been ongoing for a
few years.
Self-driving cars—Road Transport
In the United States, as of now, five states have introduced legislation and regu-
latory actions regarding autonomous transport and 15 more have pending legislative
regulations concerning the technology, according to recent news.
In Hungary, on 19th May 2016, Günther Oettinger, the European Commissioner
for Digital Economy and Society in the European Commission stated that Europe
lags behind the US regarding software research and development. The strength-
ening of cooperation of the digital market, vehicle industry and information tech-
nology has to be facilitated by creating a unified digital policy in the fields of cyber
security and new traffic regulations, and this has also become necessary with the
290 D. Tokody et al.

development of autonomous vehicles. With his proposal, the development and

production of autonomous vehicles in Europe could get closer to its competitors’
achievements. According to plans, a test track will be built in Zalaegerszeg for
testing autonomous vehicles [5].
Legislators of the European Union (http://eur-lex.europa.eu/) mention the term
of “self-driving cars” in only three documents. All of the dates of these documents
are of 2016. This clearly shows the situation of introducing autonomous road
vehicles in Europe.
15th January 2016—the opinion of the European Economic and Social
Committee on the document entitled “The effects of digitalisation on the service
industry and employment within the framework of industrial restructuring” was
issued.
“Recent years have seen major breakthroughs in the development of digital
technologies. The new and extremely powerful capabilities of digital technologies
allow for the automation of ever more tasks and occupations (e.g. the self-driving
car, the internet of Things, Industry 4.0). Additionally, digital technologies are
leading to drastic reductions in search and transaction costs and thus enabling the
development of entirely new and highly scalable business models in services (e.g.
online marketplaces and platforms including the so-called sharing economy, Uber,
Airbnb). This is encouraging the digitalisation of business models and processes in
a wide range of economic sectors. Some of them have already been addressed in
previous EESC opinions (3)” [6].
19th April 2016—The announcement of the European Committee directed
towards the European Parliament, Council, the European Economic and Social
Committee and the Committee of the Regions, concerning the digitalisation of
Europe’s industry and the advantages of fully exploiting a unified digital market:
“Stakeholders expressed a need to examine the regulatory framework for digital
innovations with a view to provide further clarity on the following: …
Autonomously acting systems such as self-driving cars or drones pose a challenge
to current safety and liability rules where a legal person is ultimately responsible.
Legal implications of the roll-out of IoT are wider than the allocation of liability as
recognised in the DSM strategy and also need to be addressed” [7].
25th May 2016—In the European Committee’s document entitled
“COMMISSION STAFF WORKING DOCUMENT Online Platforms
Accompanying the document Communication on Online Platforms and the Digital
Single Market”:
“Furthermore, online platforms are continuously changing and developing in
new directions. They are at the forefront of developments in the emerging platforms
of the future: virtual reality, connected and self-driving cars, Internet of Things, Big
Data and Artificial Intelligence” [8].
According to the documents above, it is noticeable that although the legislation
is not final, it is kept on the agenda, and based on the results of the latest research,
the digital evolution of vehicles is an important economic issue in Europe, which
triggered a new industrial revolution in the world. The building of a society based
on digital and cyber-physical systems is in progress.
An Overview of Autonomous Intelligent Vehicle Systems 291

Research of intelligent vehicles in the European Union

Autonomous vehicles use intelligent systems in order to become cost-efficiently
operated. Their environmental impact is smaller than that of traditional vehicles,
since for optimal fuel consumption during their run accelerating and decelerating
profiles are used in accordance with traffic ratios [9].
Creating an active security system is a vital issue when it comes to self-driving
vehicles. Examples of this would be the implementation of systems designed to
avoid head-on collisions and reduce risks [10–12].
Regarding vehicle development, further significant roles are played by the
systems, supervised by man, within which various driving-assistance systems are
being developed. A kind of system could be one which can adapt to the driver’s
behaviour, maintaining travel speed and signalling possible collisions (cruise
control and forward collision warning) [13].
Multiple high-volume cutting edge European projects are in progress. Such is the
AdaptIVe project, which is a 25 million Euro budget project coordinated by the
Volkswagen Group Research from 2014 to 2017. The goal of the research is to
create a highly automated vehicle. Taking into account and utilising the human
factor, a new type of cognitive integrated structure is formed by developing the
functions (28) and applications (65) of intelligent vehicles. During the research, the
handling of the systems’ complexity is a crucial aspect [14].
Several organizations are currently working on the elaboration of intelligent
transportation systems and the technological solutions related to them. This spring
in Munich at the “Trends in Automotive Radar and Impact on System Architecture
Workshop” convention, an expert in radiospectrum management, Thomas Weber at
the European Communication Office drew attention to the following:
“Regulators consider other transportation applications should share the ITS
frequencies such as Urban Rail Systems (based on an analysis in progress—looks
like it could be possible)—also perhaps relevant for future ITS applications—such
as the driverless vehicles or autonomous systems” [15].
“New ECC Decision (16)02 on rotorcraft anti-collision radars operating in 76–
77 GHz approved for publication on 4 March 2016. Be aware that 76–77 GHz,
includes automotive radars, infrastructure radars in the current regulation; it is also
used at railway level crossings or airplanes during taxiing” [15].
This claim also leads to the point of our article, which is the possibility of
interactions between railway vehicles and road vehicles at railway crossings.
Vehicle automation from conventional to intelligent
The first table shows the levels of vehicle automation, defined by the international
SAE association. From level 0 to level 5, we gradually get from a vehicle driven by
a person to a fully autonomous vehicle. With LoA 0, the driver controls the vehicle
in terms of braking, accelerating, speed, change of direction, parking, maintaining
distance and changing lanes. With LoA 1, a function-specific automation is
implemented, which helps the driver, e.g. adapting speed-maintaining automation.
With LoA 2, the vehicle driving procedures are partially automated, which further
assists driving. With LoA3, specific automated systems are cooperating with each
other. With LoA 4, the vehicle is capable of executing complex driving functions,
292 D. Tokody et al.

such as parking itself. And LoA 5 means the total automation of the vehicle, in
other words, every operating function of the vehicle is executed autonomously
without the driver.
Considering the functional description of these levels, a few important systems
from the possibilities of potential automation are listed here:
LoA 0 (driver only): Advanced Driver Assistance Systems (AEBS), Anti-lock
Braking System (ABS), Electronic Stability Control (ESC), Forward Collision
Warning (FCW), Lane Departure Warning (LDW) [14]. LoA 1 (assisted—
Automated vehicle): Adaptive Cruise Control (ACC), Park Steer Assistance, Lane
Keeping Assistance [14]. LoA 2 (partial automation—Automated vehicle): Traffic
Jam Assistant (TJA), Parking assistant [14]. LoA 3 (conditional automation—
Automated vehicle): Highway traffic jam system [14]. LoA 4 (high automation—
Automated vehicle): Parking garage pilot [14]. LoA 5 (full automation—
Autonomous vehicle): Robot Taxi [14] (Table 1).
Automatic Train Operation—Guided Transport
With the automation of the operation of vehicles [17], concerning fixed-track
transportation, significant results have been achieved, e.g. in automatic train driv-
ing, based on iterative learning [18]. In the line of the system’s automatic functions
manipulating the vehicle’s speed (using a speed profile) [19] is technologically an
already solved problem, as well as vehicle starting, stopping, standby [20], the
traffic of vehicles based on a genetic algorithm [21], collision and catch-up pro-
tection [22] and the prevention of opposite traffic situations. On the other hand,

Table 1 Levels of motor vehicle automation adapted from SAE Standard J3016 [16]
Levels of Type of Steering, Monitoring Fallback System
automation motor acceleration, driving performance of capability
vehicle deceleration environment dynamic driving (driving
operation task modes)
LoA 0 No Driver Driver Driver n/a
automated,
traditional
car
LoA 1 Drive Driver/Automatic Driver Driver Some
assistance driving
modes
LoA 2 Partial Automatic Driver Driver Some
automation driving
modes
LoA 3 Conditional Automatic Automatic Driver Some
automation driving
modes
LoA 4 High Automatic Automatic Automatic Some
automation driving
modes
LoA 5 Full Automatic Automatic Automatic All driving
automation modes
An Overview of Autonomous Intelligent Vehicle Systems 293

heavy railway automation is lagging behind its public road counterpart, since, most
probably, the application of newer technologies takes significantly more time,
because of the infrastructural costs and the calculated lifespan of the vehicles and
the infrastructure.
The average calculated lifespan of railway vehicles is 30 years or 10 million km,
of signalling and interlocking systems it is 15–50 years, while structures and
platforms must be usable for up to 100 years. In the case of automobiles, there is a
huge dispersion of lifespan periods from 300 000 up to 1 million km, and for trucks
and lorries, lifespans of 1.5–4 million km are not unusual [23–25].
There are roughly 50 (increasing) automated subway lines in the world,
including one in Budapest, Hungary. With the help of automatic operation,
timetables can reach an almost 100% accuracy. Vehicles can automatically maintain
departure and arrival times stated in the timetable, using different driving charac-
teristics. If a bigger crowd at a station would withhold the vehicle, it could work off
the delay and affect the following vehicles’ traffic itself within certain limits.
Automatic operation significantly increases the safety and energy-efficiency of
vehicle traffic. The vehicle can achieve this by using automatic speed profiles,
exceptionally accurate positioning and automatic breaking curve calculation. The
position of the vehicle is calculated by an odometer, fixed balises and using the map
of the infrastructure. The vehicles are controlled by the instructions of the inter-
locking system with the help of the CBTC system (Communications-based train
control). Traffic operators supervise the system through ATS (Automatic Train
Supervision), but the operation is fully automatic [26, 27].
IEC 62290-1 standard defines the main concepts of Urban Guided Transport
Management and Command/Control Systems. This standard includes the definition
of the levels of automation concerning fixed-track transport. These levels are shown
in the second table. GoA 1 (Manual Protected Operation, MPO: Automatic Train
Protection (ATP) with driver) ensures in a non-automatic railway operation that a
safe track path and distance between vehicles are guaranteed, and the vehicles’
partial supervision of speed is solved. GoA 2 (Semi-automated Operation Mode,
STO: ATP and Automatic Train Operation (ATO) with driver), in addition to the
previous level, is in total control of the train’s speed, braking and accelerating. GoA
3 (Driverless Train Operation, DTO) adds the ability to detect and avoid obstacles
and people on the railway. GoA 4 (Unattended Train Operation, UTO) makes a
fully automated operation possible without the need of a train attendant. This is the
operational level in which the Nr.4 subway line in Budapest operates. This level is
currently the most advanced with regard to both functionality and safety. Full
automation means that the safe traffic of trains is guaranteed, the driving of the train
is automated in accordance with the timetable and there is an active track man-
agement system detecting and avoiding objects and people falling in the path of the
train. Moreover, supervision of the transport of people, automated closing of the
passenger doors and the guarantee of safe start-up conditions are all ensured
[28, 29] (Table 2).
Automation considerably improves the competitiveness of the railway sector.
The development of heavy railway automation and the installation of systems, like
294 D. Tokody et al.

Table 2 Grades of automation by UITP [30]

Grades of Type of train Setting Stopping Door Operation in
automation operation train in train closure event of
motion disruption
GoA 1 ATP with Driver Driver Driver Driver
driver
GoA 2 ATP and Automatic Automatic Driver Driver
ATO with
driver
GoA 3 Driverless Automatic Automatic Train Train attendant
attendant
GoA 4 UTO Automatic Automatic Automatic Automatic

ETCS (European Train Control System), ERTMS (European Rail Traffic

Management System), GSM-R is an important step in the sector. At the same time,
establishing ETCS Level 4 is inevitable, in which trains (not only subway lines) can
operate automatically without a driver. Fixed-track, ground-level transport could be
a good alternative to road traffic, when you consider the differences in capacity in
both passenger and cargo transport, or the possibility of energy-saving provided by
automated operation [31].

3 Technological Overview of the Development

of Intelligent Vehicle Systems

3.1 Definitions/Terminology of Intelligent Vehicle

Technology

Co-operative intelligent transport system (C-ITS): “In the previous section a new
concept has emerged, namely the concept of ‘Intelligent Transport Systems’ (ITS).
It refers to the endeavour to develop the integrated operation of various transport
structures by applying the research results of interdisciplinary fields, for example by
using the infocommunication technologies. Logistic hubs are also in the focus of
this endeavour. Main aims of Intelligent Transport Systems are to find more
environmentally friendly methods of transport, to implement an efficient transport
system and to enhance transportation safety” [32].
Intelligent infrastructure: “Intelligent infrastructure can be defined as an inte-
grated system which includes the complete traditional intra- and interurban
infrastructures, collects data of them (with the help of sensors), and, by evaluating
these data, improves the efficiency and safety of operation. It also optimizes and
maintains the operation of the city, and supports environmental protection efforts.
Based on the collected data, this system is able to provide help for humans to
prevent accidents. This can be achieved by carrying out an early diagnosis of the
An Overview of Autonomous Intelligent Vehicle Systems 295

problems in the system based on the evaluation of the Big Data collected and
analysed about the system. Furthermore, the ‘learning ability’ of the system helps
to recognise such schemes which can lead to abnormal operation. The intelligent
infrastructure provides the basis for the smart city. The intelligent infrastructure
is not limited to the city centre only. This system also connects different smart
cities” [33].
Smart sustainable city: “A smart sustainable city (SSC) is an innovative city that
uses information and communication technologies (ICTs) and other means to
improve quality of life, efficiency of urban operation and services, and competi-
tiveness, while ensuring that it meets the needs of present and future generations
with respect to economic, social and environmental aspects” [34].
Smart mobility: “The concept of mobility is used for the moving and travelling
of people.” “Smart mobility is one of the main characteristics of the smart city.”
“Mobility: Safe and green mobility is possible because of intelligent vehicles and
coordinated traffic management systems with the help of distributed real-time sit-
uation awareness and solution finding” [35].
Intelligent vehicle: A system of robotic applications to collect information on the
position, kinematics and dynamics of the vehicle, the state of the environment and
the state of the driver and the passenger, to assess such information and make
decisions based on it. It is capable of duplex communication with roadside
infrastructure and other vehicles, to use digital map applications and satellite
positioning systems, it has an active internet connection and its own physical
address [36].
Self-driving car (Autonomous car): A vehicle which is able to perform the
driving tasks without the intervention of the driver, with a high degree of safety, to
recognise any obstacles in its environment, and to stop before or go round such
obstacles. It can communicate with the surrounding infrastructure and with various
other vehicles. It has internet access and route design capable of online change
management. It has special driving patterns and ensures its energy supply in an
autonomous way. It can park autonomously [37–39].
V2V communication (VANET—Vehicular ad hoc network): It is an ad hoc
mesh network-based wireless data transmission method for inter-vehicle commu-
nication. Its purpose is to enable vehicles to share information about their position,
speed, direction or about any dangers in traffic, based on which they can take the
necessary preventive measures, i.e. they can slow down or stop in time. It is one of
the dedicated short-range communications technologies [40–43].
V2I and I2V communication: It ensures information share between the infras-
tructure and the vehicles approaching a crossroads. It can also provide data for a
parking vehicle to start its route, or send speed instructions to vehicles. It can also
be used for communication between railway level crossings and vehicles [44].
Further applications: “Red Light Violation Warning, Curve Speed Warning, Stop
Sign Gap Assist, Reduced Speed Zone Warning, Spot Weather Information
Warning, Stop Sign Violation Warning, Railroad Crossing Violation Warning, and
Oversize Vehicle Warning” [44].
296 D. Tokody et al.

Vehicle-to-Everything (V2X): A system including several communication

methods: Vehicle-to-Infrastructure (V2I), Vehicle-to-vehicle (V2V), Vehicle-
to-Pedestrian (V2P), Vehicle-to-device (V2D), Vehicle-to-grid (V2G) [45].
GSM-R communication: The communication system of fixed-track vehicles for in-
frastructure-to-vehicle communication. Global System for Mobile Communications—
Railway [46].
IoT based vehicle technology (vehicle-to-cloud communication, cloud technol-
ogy for vehicles system): The technology of the Internet of Things connects various
devices in the virtual space. It can transfer a large amount of data to extra-vehicular
applications. It takes advantage of cloud technology to connect vehicles to the
infrastructure, other vehicles, intelligent buildings and the smart grid for a more
efficient resource management [47]. It can provide the basis for a toll system [48].
Smart grid and electric vehicles: The growing number of electric vehicles causes
a considerable increase in electric energy consumption in the world. Meanwhile,
some research proves that, besides their consumption, the energy producing and
storing capacity of vehicles can also change the characteristics and operating phi-
losophy of the supply network in a considerable way. It is based on the fact that the
parking and charging time of these vehicles are certainly longer than their daily
running time [49–51].
Network of everything, networked society: “Cities in themselves can also be
described as networks, or the aggregate of systems and subsystems. The individual
elements comprise a continuously changing artificial system, which functions like a
living organism, and its technological complexity has significantly increased over
the last decades” [52]. The network elements are the intelligent vehicles too.
Cyber-physical systems (CPS): “Cyber-Physical Systems are such systems
which connect the virtual world and the physical world with the help of embedded
systems and infocommunication systems.” It is a key element of vehicle automa-
tion, facilitating the digitalisation of vehicles [52].
Micro-electromechanical systems for passenger safety in automotive vehicles:
As a result of the technological revolution of the 21st century, the size of the
devices used in robotic technology has been reduced considerably. The use of
silicon-based microelectronic and michromechanic constructions in the vehicle
industry has now become everyday practice. Micro-sized electro-mechanical
devices are capable of detection, control, intervention and influence at a micro level.
MEMS technology is used in the following systems: automotive airbag sensors,
internal navigation sensors, air conditioning compressor sensors, brake force sen-
sors and suspension control accelerometers, fuel level and vapour pressure sensors,
“Intelligent” tyres [53].
Multi-agent system (MAS): Its sensor agents are able to detect the environment,
while its intervention agents are able to change it. Multi-agent systems can model
the complex processes of nature, as its individual agents act in co-operation with
each other in a dynamically changing environment, while performing their specific
tasks in the system and creating common knowledge through interactive
information-based communication [54, 55].
An Overview of Autonomous Intelligent Vehicle Systems 297

Inter-Vehicle Communication (IVC): “Intervehicle communication

(IVC) networks, a subclass of mobile ad hoc networks (MANETs), have no fixed
infrastructure and instead rely on the nodes themselves to provide network func-
tionality. However, due to mobility constraints, driver behaviour, and high
mobility, IVC networks exhibit characteristics that are dramatically different from
many generic MANETs. This paper elicits these differences through simulations
and mathematical models and then explores the impact of the differences on the
IVC communication architecture, including important security implications” [56].
Automated Transport Systems: “An Automated Transport System (ATS) is an
innovative holistic mobility concept, where all its different elements (i.e. vehicle,
travellers, public transport, infrastructure, operations and control) are capable of
self-organizing and operating at an “automated” manner, addressing in real time the
needs of all and each participant of a specific traffic scenario, applying different
levels of automation and supporting all transport modes for both passenger and
freight” [57], © Original definition, ICCS, 2013.
In addition to their definitions, Fig. 1 aims to show the relationships between the
above concepts revealing the interconnections that can be important from the point
of our research. As it has been illustrated in Fig. 1, a smart, sustainable city has six
primary elements [58], and one of these elements is smarty mobility. Smart mobility
requires the creation of an intelligent cyber-physical infrastructure, which involves
the technologies of telemetrics, cybernetics, artificial intelligence, telecommunica-
tion and infocommunication. From a holistic aspect, the development of intelligent
transport systems also includes the various methods of transport in the whole
spectrum of transportation.
The autonomous systems which have already been developed are characterised
by a communication-based structure. Autonomous intelligent cars communicate
with each other through short-range communication (several metres) or long-range
communication (50–80 m). These communication solutions are enough, for
example, for the communication of a tram or a car. However, they are not suitable
for the communication of a high-speed train. Further technological developments
are needed to find a solution to this problem.

4 Failure Mode Effects Analysis

Definitions:
• Failure mode: the type of failure.
• Failure effect: Effects and consequences caused by the failure.
• Failure cause: Defining the exact cause of the failure.
• Failure extent: The gravity of the failure’s consequences.
• Frequency: The frequency of the failure’s occurrence.
• Recognisability: The efficiency with which the failure can be detected.
• Measures: Actions preventing the recurrence of a failure.
298 D. Tokody et al.

Fig. 1 The relationship between Intelligent Transportation Systems and intelligent vehicles
(revised figure) [59]

The aim of Sect. 4 is to realise complex, intelligent vehicle systems, which can
collectively and automatically handle vehicles equipped with rubber wheels (cars,
trucks, motorcycle, etc.) or with iron wheels (heavy railway, subway, suburban
railway, tram, etc.). These systems could still be called “utopian”, but a future goal
is to make them reality. It can be clearly seen that this is an incredibly complex
system, therefore, regarding the FMEA analysis, we mainly concentrated on the
main hazards, system components and errors, and less on the errors of sub-systems
(Table 3).
Table 3 Failure mode effects analysis
Number Failure mode Failure effect Sever Cause Occur Detect RPN Provisions
1 Turning off Speeding irregular 10 Free provision 10 10 1000 Restricted overwrite options
human direction over the
intervention system
2 Hacker attack Taking over the vehicle’s 10 System is not 6 10 600 Testing
control/incorrect data sufficiently
communication encrypted
3 Faulty code Speeding collision hazard 10 Human 5 10 500 Multiple quality controls
mistake and tests
4 Weather Change in braking distance 10 – 5 10 500 Preparing the system for
anomalies data loss extreme weather
5 Wheeled vehicle Positioning loss 10 Technical 4 9 360 In case of signal loss,
signal loss failure, change to manual control
shielding
6 Iron wheeled Positioning loss 10 Technical 4 9 360 In case of signal loss,
vehicle signal failure, change to manual control
loss shielding
An Overview of Autonomous Intelligent Vehicle Systems

7 Distinguished Distinguished vehicles are 10 Incorrect 7 6 420 Indicating distinguished

vehicle handling not integrated to the handling of vehicles to the system
system, therefore increased distinguished
accident risk vehicle
8 Oversized Oversized vehicle badly 10 Incorrect 4 8 320 Data handling of oversized
vehicle handling positioned delimitation handling of vehicles
thus creating a dangerous oversized
situation vehicle’s
specialties
(continued)
299
Table 3 (continued)
300

Number Failure mode Failure effect Sever Cause Occur Detect RPN Provisions
9 Wheeled Incorrect positioning speed 10 Technical 5 9 450 Using a proper time limit,
vehicle’s delayed setting failure, after which the data loss is
data shielding considered as delayed
sending/receiving sending/receiving
10 Iron wheeled Incorrect positioning speed 10 Technical 5 9 450 Using a proper time limit,
vehicle’s delayed setting failure, after which the data loss is
data shielding considered as delayed
sending/receiving sending/receiving
11 Sending incorrect Incorrect positioning and 10 – 2 8 160 –
data to vehicles inaccurate speed setting
12 Wheeled vehicle Road conditions, traffic 10 – 3 9 270 –
sensor failure barriers and loss of sensing
other vehicles
13 Iron wheeled Traffic barriers and loss of 10 – 3 9 270 –
vehicle sensor sensing other vehicles
failure
14 Technical failure Termination of the 10 Poorly 1 10 100 Strict in-process quality
operation of sub-systems, developed control
system shutdown physical
parameters
15 Unexpected Unexpected appearance of 8 Unexpected 5 10 400 –
effects of passive pedestrians, animals, effects of
components coercion to emergency passive
manoeuvre components
D. Tokody et al.
An Overview of Autonomous Intelligent Vehicle Systems 301

5 Possibilities to Increase Safety by Using Intelligent

Systems

Active vehicle safety systems, which are based on intelligent communication and
used at road and railway level crossing and in railway infrastructure, have the
following elements:
• Intelligent infrastructure agent (traffic environment, traffic lights, cameras, sig-
nallers, gates, switches and crossings, sensor networks, etc.).
• Intelligent railway vehicle agent.
• Intelligent road vehicle agent.
The agent is able to detect its environment and interact with it for its own
interests through its interventive capacity. The autonomous behaviour of intelligent
agents depends on their freedom to make decisions based their accumulated
knowledge.
The agent-based conception of intelligent transport systems means that an
ambient sensor system continuously monitors the state of the transport system by
sensing the various transport infrastructure elements and transport system units
(vehicles, participants in traffic). The types of sensors include radars, infra gates,
opening sensors, pressure sensors, stretch sensors, acceleration sensors, etc. The
agents are continuously capable of intervening, as without intervention or reaction
none of the autonomous elements could perform their tasks. Not only the transport
infrastructure itself, but the whole environment could be intelligent.
Vehicle and fixed-track vehicle, or intelligent vehicle agents include the fol-
lowing: communication module, vehicle intervention/control/braking module,
information assessing and decision making unit, interface for on-board systems,
module for collecting and storing information from the authorities, black box, etc.
Adaptive running properties and timetable subsystem: fixed-track vehicles run
according to a well-defined timetable, but various traffic situations can dynamically
modify it. This dynamic timetable is continuously revised and corrected by the
system, and it is sent to the road vehicles within a certain zone. The running
properties of road vehicles will also be sent to fixed-track vehicles through a
wireless communication system. Each vehicle agent has their own running prop-
erties and timetable.
The key element of the system is the wireless communication network which
helps to transfer the information created by the agents. Possible types of the
wireless communication system by the parties of communication:
• Train to train communication,
• Train to infrastructure and infrastructure to train communication,
• Train to road vehicle and vehicle to train communication,
• Vehicle to infrastructure and infrastructure to vehicle communication.
In order to achieve autonomous transport, a database of intelligent infrastructure
maps (map of railway lines, road maps) is required, which can be partly found in all
302 D. Tokody et al.

vehicle agents, or retrieved from fix infrastructure agents. Maps are modified and
corrected by the agents.
The above-described system is demonstrated by a system composition in Fig. 2,
which includes the following numbered elements: 1 traditional and high-speed
passenger trains, 2 freight trains, 3 road vehicles transporting goods, 4 cars and
vehicles using distinguishing signals, 5 railway gate equipment, 6 communication,
7 railway tracks.
Messages of inter-vehicle communication: speed, exact position, direction of
movement and the priority of the vehicle. Communication messages of the fixed
infrastructure: operability and state (e.g. gate is open or closed), while the infras-
tructure can also give speed restrictions, brake warning, or immediate stop
instruction.
The messages will be received by all vehicles within the defined zone around the
railway gate. Zones will be defined by considering the longest general braking
distance of the vehicles, both in the case of the road and the railway vehicles.
Long-distance communication, within 50–80 m, has already been realised between
cars.
The system recognising adaptive traffic situations gives greater priority to
fixed-track vehicles. It also has an emergency plan for unavoidable collisions. The
virtual protected environment surrounding the vehicle depends on the speed and the
direction of approaching or distancing. The system makes it possible to use the
radar and the camera together. The images of the cameras can be seen in the
vehicles, but intervention may only happen by the automatic assessment of other
sensors and image processing. All vehicles contain a data recorder unit given by the
authorities (a black box—EDR—event data recorder).

Fig. 2 Intelligent communication-based active vehicle protection system used at road and railway
level crossings and in railway infrastructure (figure by author)
An Overview of Autonomous Intelligent Vehicle Systems 303

6 Conclusion

In conclusion, more than 92% of accidents can be directly attributed to human

errors. This is also true for the accidents involving road vehicles and fixed-track
vehicles. There is hardly any intersection between road transport and fixed-track
transport developments. The automation endeavours presented in our article
envisage a uniform autonomous intelligent operation for both road transport and
railway vehicles. We see the opportunity to increase safety in the application of
intelligent systems. Road and railway level crossings are the most obvious hazards.
According to our suggestion, it is possible to implement an active vehicle safety
system used in railway infrastructure at road and rail level crossings based on
intelligent communication. It can also be part of intelligent infrastructure as we
defined it, and it relates to smart mobility, one of the basic pillars of smart city.

Acknowledgement The research on which the publication is based has been carried out within
the framework of the project entitled “The Development of Intelligent Railway Information and
Safety Systems”. This research has been realised by using the resources of the National Talent
Programme, Grant Scheme for the Nation’s Young Talents (Application number: NTP-NFTÖ-
16-0582) and the support of the Human Resource Support Office and the Ministry of Human
Resources.

References

1. Statistical Yearbook of Traffic Accidents (2014) [Online]. https://www.ksh.hu/docs/hun/xftp/

idoszaki/baleset/baleset14.pdf. Accessed 10 Aug 2016
2. Directive 2010/40/EU, Intelligent Transport Systems in the field of road transport, as well as
framework for interfaces with other transport modes [Online]. http://eur-lex.europa.eu/legal-
content/HU/TXT/PDF/?uri=CELEX:32010L0040&from=EN. Accessed 10 Aug 2016
3. Mester Gy (2010) Service robots, VMTT Konferencia, Novi Sad, Serbia. ISBN
978-86-88077-02-6 (in Hungarian)
4. World’s first electric road opens in Sweden [Online]. https://www.scania.com/group/en/
worlds-first-electric-road-opens-in-sweden/. Accessed 10 Aug 2016
5. Előd F (2016) According to Orbán and Varga the autonomous car-manufacturing in the future
of the Hungarian industry [Online]. http://index.hu/gazdasag/2016/05/19/orban_es_varga_
szerint_az_onvezeto_autogyartas_a_magyar_ipar_jovoje/. Accessed 12 Aug 2016 (in
Hungarian)
6. Opinion of the European Economic and Social Committee on ‘The effects of digitalisation on
the services sector and employment in relation to industrial change’(exploratory opinion)
(2016/C 013/24) [Online]. http://eur-lex.europa.eu/legal-content/EN/TXT/?qid=14702584
24676&uri=CELEX:52015AE0765. Accessed 12 Aug 2016
7. Communication from the commission to the European Parliament, The Council, The
European Economic and Social Committee and the Committee of the Regions Digitising
European Industry Reaping the full benefits of a Digital Single Market {SWD(2016) 110
final}, Brussels [Online]. http://eur-lex.europa.eu/legal-content/EN/TXT/?qid=14702584
24676&uri=CELEX:52016DC0180. Accessed 12 Aug 2016
8. Commission staff working document, Online Platforms Accompanying the document
Communication on Online Platforms and the Digital Single Market {COM(2016) 288 final},
304 D. Tokody et al.

Brussels [Online]. http://eur-lex.europa.eu/legal-content/EN/TXT/?qid=1470258424676&

uri=CELEX:52016SC0172. Accessed 12 Aug 2016
9. Urmson C, Whittaker W (2008) Self-driving cars and the urban challenge. IEEE Intell Syst 23
(2):66–68. doi:10.1109/MIS.2008.34. ISSN 1541-1672, URL: http://ieeexplore.ieee.org/
stamp/stamp.jsp?tp=&arnumber=4475861&isnumber=4475846
10. Kim T, Jeong HY (2010) Crash probability and error rates for head-on collisions based on
stochastic analyses. IEEE Trans Intell Transp Syst 11(4):896–904. doi:10.1109/TITS.2010.
2053536. ISSN 1524-9050. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=
5504226&isnumber=5635426
11. Sugimoto M et al (2000) Realization of head-on collision warning system at
intersections-DSSS: driving safety support systems. In: Intelligent vehicles symposium, 2000.
IV 2000. Proceedings of the IEEE, Dearborn, MI, 2000, pp 731–735. doi:10.1109/IVS.2000.
898436. ISBN 0-7803-6363-9. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=
898436&isnumber=19432
12. El-Shamy EF (2010) Head-on collisions of two electrostatic solitary waves in electron–
positron–ion plasmas. IEEE Trans Plasma Sci 38(4):909–914. doi:10.1109/TPS.2010.
2042465. ISSN 0093-3813. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=
5433006&isnumber=5444580
13. Wang J et al (2013) An adaptive longitudinal driving assistance system based on driver
characteristics. IEEE Trans Intell Transp Syst 14(1):1–12. doi:10.1109/TITS.2012.2205143.
ISSN 1524-9050. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6236181&is
number=6470717
14. Kelsch J (2016) Human-Vehicle Integration in EU-AdaptIVe, ITS World Congress,
Bordeaux, France [Online]. https://www.adaptive-ip.eu/files/adaptive/content/downloads/
Deliverables%20%26%20papers/Kelsch_AdaptIVe%20SP3%202015%20-%20ITSCongress.
pdf. Accessed 25 Aug 2016
15. Weber T (2016) ECO, spectrum management: activities in CEPT (European Spectrum
Regulation) for applications in transport and traffic telematics including sensors, trends in
automotive radar and impact on system architecture workshop, March 14–17. Munich,
Germany
16. SAE International (2016) Automated driving levels of driving automation are defined in New
Sae International Standard J3016 [Online]. http://www.sae.org/misc/pdfs/automated_driving.
pdf. Accessed 25 Aug 2016
17. Venkateswaran KG et al (2015) Impact of automation on the capacity of a mainline railway: a
preliminary hypothesis and methodology. In: 2015 IEEE 18th international conference on
intelligent transportation systems, Las Palmas, 2015, pp 2097–2102. doi:10.1109/ITSC.2015.
339. ISSN 2153-0009. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7313431
&isnumber=7312804
18. Wang Y et al (2008) A novel automatic train operation algorithm based on iterative learning
control theory. In: IEEE international conference on service operations and logistics, and
informatics, 2008. IEEE/SOLI 2008, Beijing, pp 1766–1770. doi:10.1109/SOLI.2008.
4682815. ISBN 978-1-4244-2012-4. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&ar
number=4682815&isnumber=4682741
19. Dominguez M et al (2012) Energy savings in metropolitan railway substations through
regenerative energy recovery and optimal design of ATO speed profiles. IEEE Trans Autom
Sci Eng 9(3):496–504. doi:10.1109/TASE.2012.2201148. ISSN 1545-5955. http://ieeexplore.
ieee.org/stamp/stamp.jsp?tp=&arnumber=6213573&isnumber=6226882
20. Zuo J et al (2010) Feedback control of pneumatic brake of urban railway train under ATO
mode. In: 2010 international conference on E-Product E-Service and E-Entertainment
(ICEEE), Henan, 2010, pp 1–4. doi:10.1109/ICEEE.2010.5660567. ISBN
978-1-4244-7159-1. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=5660567&is
number=5660092
21. Chang CS, Sim SS (1997) Optimising train movements through coast control using genetic
algorithms. In: IEE proceedings—electric power applications, vol 144, no 1, pp 65–73.
An Overview of Autonomous Intelligent Vehicle Systems 305

doi:10.1049/ip-epa:19970797. ISSN 1350-2352. http://ieeexplore.ieee.org/stamp/stamp.jsp?

tp=&arnumber=583371&isnumber=12657
22. Liu J et al (2013) An analysis of BeiDou Navigation Satellite System (BDS) based positioning
for train collision early warning. In: Intelligent vehicles symposium (IV), 2013 IEEE, Gold
Coast, QLD, 2013, pp 1065–1070. doi:10.1109/IVS.2013.6629607. ISSN 1931-0587. http://
ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6629607&isnumber=6629437
23. Skillingberg M, Green J (2007) Aluminum applications in the rail industry. Light Metal
Age-Chicago 65(5):8
24. Jun HK, Kim JH (2007) Life cycle cost modeling for railway vehicle. In: ICEMS.
international conference on electrical machines and systems, Seoul, 2007, pp 1989–1994.
ISBN 978-89-86510-07-2. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=4412
054&isnumber=4411938
25. Baumgartner JP (2001) Prices and costs in the railway sector. Laboratoire d’Intermodalité des
Tansports Et de Planification, École Polytechnique Fédérale De Lausanne
26. Dominguez M et al (2012) Energy savings in metropolitan railway substations through
regenerative energy recovery and optimal design of ATO speed profiles. IEEE Trans Autom
Sci Eng 9(3):496–504. doi:10.1109/TASE.2012.2201148. ISSN 1545-5955. http://ieeexplore.
ieee.org/stamp/stamp.jsp?tp=&arnumber=6213573&isnumber=6226882
27. Zhu L et al (2012) Handoff performance improvements in MIMO-enabled communication-
based train control systems. IEEE Trans Intell Transp Syst 13(2):582–593. doi:10.1109/TITS.
2012.2188288. ISSN 1524-9050. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=
6183515&isnumber=6205705
28. Darai L (2013) Metrók, metró biztonsága, oktatási vázlat, slide show [Online]. http://www.
kjit.bme.hu/images/stories/targyak/metro/automatiz_goa_5r.pdf. Accessed 10 Jan 2016
29. Oh S, Yoon Y, Kim Y (2012) Automatic train protection simulation for radio-based train
control system. In: 2012 international conference on information science and applications,
Suwon, 2012, pp 1–4. doi:10.1109/ICISA.2012.6220965. ISBN 978-1-4673-1402-2. http://
ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6220965&isnumber=6220919
30. International association of public transport: metro automation facts, figures and trends, p 1,
2012, Brüsszel [Online]. http://metroautomation.org/wp-content/uploads/2012/12/Automated_
metros_Atlas_General_Public_2012.pdf. Accessed 26 Aug 2016
31. Rail automation—reaching out across Europe, 6th August 2016 [Online]. http://www.
railengineer.uk/2016/08/06/rail-automation-reaching-out-across-europe/. Accessed 26 Aug
2016
32. Tokodi D, Schuster G, Ihász J (2014) SMART Rail technológiák lehetőségei, az intelligens
vasúti hálózatok kialakításának kérdései. Vezetékek Világa: magyar vasúttechnikai szemle 19
(2):11–15
33. Tokody D, Schuster Gy (2015) I2—Intelligent Infrastructure. In: Schmidt P et al
(ed) Reviewed proceedings fifth international scientific videoconference of scientists and
PhD. students or candidates: trends and innovations in E-business, education and security,
129 pp, Bratislava, Szlovákia. University of Economics in Bratislava, pp 121–128. ISBN
978-80-225-4191-6
34. International Telecommunication Union, smart sustainable cities: an analysis of definitions,
p 21 [Online]. https://www.itu.int/en/ITU-T/focusgroups/ssc/Documents/Approved_
Deliverables/TR-Definitions.docx. Accessed 21 Dec 2015
35. Tokody D, Holicza P, Schuster G (2016) The smart mobility aspects of intelligent railway. In:
2016 IEEE 11th international symposium on applied computational intelligence and
informatics (SACI), pp 323–326. IEEE
36. Trebuňa F et al (2012) Modelling of mechanical and mechatronics systems intelligent vehicles
as the robotic applications. Procedia Eng 48:105–114. ISSN 1877-7058. http://dx.doi.org/10.
1016/j.proeng.2012.09.492
37. Milanés V et al (2010) Clavileño: Evolution of an autonomous car. In: 2010 13th international
IEEE conference on intelligent transportation systems (ITSC), Funchal, 2010, pp 1129–1134.
306 D. Tokody et al.

doi:10.1109/ITSC.2010.5624971. ISSN 2153-0009. http://ieeexplore.ieee.org/stamp/stamp.

jsp?tp=&arnumber=5624971&isnumber=5624963
38. Jo K, Sunwoo M (2014) Generation of a precise roadway map for autonomous cars. IEEE
Trans Intell Transp Syst 15(3):925–937. doi:10.1109/TITS.2013.2291395. ISSN 1524-9050.
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6680774&isnumber=6823199
39. Pozna C, Antonya C (2016) Issues about autonomous cars. In: 2016 IEEE 11th international
symposium on applied computational intelligence and informatics (SACI), Timisoara,
Romania, 2016, pp 13–18. doi:10.1109/SACI.2016.7507360. ISBN 978-1-5090-2379-0.
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7507360&isnumber=7507333
40. Uhlemann E (2015) Introducing connected vehicles [connected vehicles]. IEEE Vehicular
Technol Mag 10(1):23–31. doi:10.1109/MVT.2015.2390920, ISSN 1556-6072. http://
ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7047294&isnumber=7047264
41. Kaempchen N et al (2009) Situation assessment of an autonomous emergency brake for
arbitrary vehicle-to-vehicle collision scenarios. IEEE Trans Intell Transp Syst 10(4):678–687.
doi:10.1109/TITS.2009.2026452. ISSN 1524-9050. http://ieeexplore.ieee.org/stamp/stamp.
jsp?tp=&arnumber=5161310&isnumber=5338639
42. Tokody D (2016) Smart mobility. In: Enikő B (ed) Proceedings of the XXI-th international
scientific conference of young engineers, 452 pp, Kolozsvár, Románia. Erdélyi
Múzeum-Egyesület (EME), pp 401–404
43. Sukuvaara T, Nurmi P (2009) Wireless traffic service platform for combined
vehicle-to-vehicle and vehicle-to-infrastructure communications. IEEE Wireless Commun
16(6):54–61. doi:10.1109/MWC.2009.5361179. ISSN 1536-1284. http://ieeexplore.ieee.org/
stamp/stamp.jsp?tp=&arnumber=5361179&isnumber=5361167
44. Harding J et al (2014) Vehicle-to-vehicle communications: readiness of V2V technology for
application (Report No. DOT HS 812 014). National Highway Traffic Safety Administration,
Washington, DC
45. Seo H et al (2016) LTE evolution for vehicle-to-everything services. IEEE Commun Mag 54
(6):22–28. doi:10.1109/MCOM.2016.7497762. ISSN 0163-6804. http://ieeexplore.ieee.org/
stamp/stamp.jsp?tp=&arnumber=7497762&isnumber=7497754
46. Tokody D, Maros D, Schuster Gy, Tiszavölgyi Zs. (2016) Communication-based intelligent
railway—implementation of GSM-R system in Hungary. In: 2016 IEEE 14th international
symposium on applied machine intelligence and informatics (SAMI), Herlany, pp 99–104.
doi:10.1109/SAMI.2016.7422989
47. Hank P et al (2013) Automotive ethernet: in-vehicle networking and smart mobility. In:
Design, automation & test in Europe conference & exhibition (DATE), Grenoble, France,
2013, pp 1735–1739. doi:10.7873/DATE.2013.349. ISSN 1530-1591. http://ieeexplore.ieee.
org/stamp/stamp.jsp?tp=&arnumber=6513795&isnumber=6513446
48. Cvijić B et al (2016) Cloud based web application supporting vehicle toll payment system. In:
2016 5th Mediterranean conference on embedded computing (MECO), Bar, Montenegro,
2016, pp 489–492. doi:10.1109/MECO.2016.7525700. ISBN 978-1-5090-2220-5. http://
ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7525700&isnumber=7525673
49. Ipakchi A, Albuyeh F (2009) Grid of the future. IEEE Power Energy Mag 7(2):52–62. doi:10.
1109/MPE.2008.931384. ISSN 1540-7977. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=
&arnumber=4787536&isnumber=4787510
50. Deilami S et al (2011) Real-time coordination of plug-in electric vehicle charging in smart
grids to minimize power losses and improve voltage profile. IEEE Trans Smart Grid 2
(3):456–467. doi:10.1109/TSG.2011.2159816. ISSN 1949-3053. http://ieeexplore.ieee.org/
stamp/stamp.jsp?tp=&arnumber=5986769&isnumber=5993785
51. Sortomme E et al (2011) Coordinated charging of plug-in hybrid electric vehicles to minimize
distribution system losses. IEEE Trans Smart Grid 2(1):198–205. doi:10.1109/TSG.2010.
2090913. ISSN 1949-3053. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=
5664815&isnumber=5715606
An Overview of Autonomous Intelligent Vehicle Systems 307

52. Tokody D, Schuster G (2016) Driving forces behind smart city implementations—The Next
Smart Revolution, manuscript
53. Prime Faraday Technology Watch—January 2002: an introduction to MEMS.
ISBN 1-84402-020-7. http://www.lboro.ac.uk/microsites/mechman/research/ipm-ktn/pdf/
Technology_review/an-introduction-to-mems.pdf
54. Tokody D, Papp J, Schuster Gy (2015) The challenges of the intelligent railway network
implementation: Initial thoughts from Hungary. In: Gogolák L, Fürstner I (eds) Proceedings
of the 3rd international conference and workshop mechatronics in practice and education—
MECHEDU 2015, Szabadka, Szerbia, 2015.05.14–2015.05.16. Subotica Technical College
of Applied Sciences, pp 179–185. ISBN 978-86-918815-0-4
55. Mesterséges Intelligencia Almanach [Online]. http://mialmanach.mit.bme.hu/aima/ch01.
Accessed 21 Dec 2015
56. Blum JJ et al (2004) Challenges of intervehicle ad hoc networks. IEEE Trans Intell Transp
Syst 5(4):347–351. doi:10.1109/TITS.2004.838218. ISSN 1524-9050. http://ieeexplore.ieee.
org/stamp/stamp.jsp?tp=&arnumber=1364012&isnumber=29884
57. Lytrivis P (2015) A holistic approach for automated transport systems. In: iMobility forum
plenary meeting, 21 October 2015, Brussels
58. Cohen B (2015) The smartest cities in the world 2015: methodology [Online]. http://www.
fastcoexist.com/3038818/the-smartest-cities-in-the-world-2015-methodology. Accessed 20
Dec 2015
59. Siergiejczyk M (2015) Communication architecture in the chosen telematics transport
systems. [Online]. http://cdn.intechopen.com/pdfs-wm/37575.pdf. Accessed 21 Dec 2015
Software Reliability of Complex Systems
Focus for Intelligent Vehicles

György Schuster, Daniel Tokody and Imre János Mezei

Abstract Using software became a part of our everyday life, in the last few dec-
ades. Software is widely used in areas, such as national defence, aeronautics and
astronautics, medicine or even transport. There are 100 million lines of codes in a
modern high-end car’s engine control unit. In comparison, the Space Shuttle needs
400 000, the F22 fighter jet needs less than 2 million, the Boeing 787 airplane needs
14 million and the Facebook needs more than 60 million lines of codes to function.
Even a smaller error can lead to devastating consequences in safety-critical systems,
such as those operating in vehicles. There have been several examples in recent
years, when an automotive recall was necessary due to dangerous software, and
there were cases when these errors presumably caused fatal accidents. Definition of
software reliability is the error-free working probability of software for a specified
period of time under well-defined environment. Usage of software is inevitable. It
can be found in every vehicle to control almost everything. Therefore software can
be considered as a critical success factor and it has a strong effect on the reliability
of the whole system. The software systems are getting more and more complex.
Known fact is a more complex system has more possibility to have errors. The most
difficult problem is that the traditional methods of reliability cannot be used. For
example fatigue and wearing of mechanical parts or features of lubricant systems
can be calculated quite well, since we have enough prior knowledge on their
features. Unfortunately, in case of software systems this knowledge is missing. This
paper deals with the question of software reliability. In the first part it lists the

G. Schuster (&)
D. Tokody
I.J. Mezei

Doctoral School on Safety and Security Sciences, Óbuda University,
Budapest, Hungary
e-mail: schuster.gyorgy@kvk.uni-obuda.hu
D. Tokody
e-mail: tokodi.daniel@bgok.hu
I.J. Mezei
e-mail: mezei.janos.imre@gmail.com

© Springer International Publishing AG 2017 309

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_28
310 G. Schuster et al.

problems and the second part gives some mathematical issues to calculate working
probability.

Keywords Software reliability
Complex systems
Safety critical system

Intelligent vehicles

1 Introduction

1.1 Cases

On 9 May 2015, an Airbus A400M Atlas cargo plane on a test flight crashed near
Seville, Spain. Four of the six aircraft crew were killed and the remaining two were
seriously injured. Three of the aircraft’s four engines failed during the A400M’s
departure from Seville. The crash was caused by software [1].
The US air safety authority issued a warning and maintenance order 1 May,
2015 over a software bug that causes a complete electrical shutdown of Boeing’s
787 Dreamliner. The software bug was found in the plane’s generator-control units
[2, 3].
Non-professionals would think once a software works, it will not become faulty.
Reality, however, differs. Since software—like human controllers—makes deci-
sions; its operation can also be confused by environmental effects. A software,
which worked well for the Ariane 4 rocket, caused the explosion of Ariane 5 on its
first voyage. The cause is simple; the overflow of a variable [4].

1.2 Definitions

Software: the whole of programs and data. The part of the system that is not
tangible.
Software reliability: the probability of failure-free software operation for a
specified period of time in a specified environment [5, 6].
Failure: unexpected software behaviour perceived by the user.
Fault: the software characteristic caused by failure.

1.3 Circumstances that Have an Effect on the Reliability

of the Software

• Human factor: Experience and ability to faultless work of people writing the
program and providing the data. The skilfulness of those doing the tests is
equally important [5, 6].
Software Reliability of Complex Systems Focus … 311

• Software development model: Agile and Scrum software development practices

should be avoided in the development of safety-critical systems [5, 6].
• Aptitude of management: unfit management is a recurrent problem. Managers
are often not competent in the given field (with some exceptions), and what they
expect is many times senseless or unachievable [5, 6].
• Faultless development environment: In many cases a faultless source code does not
work properly because of the faults of the compiler or the software library [5, 6].
• Faultless test environment: testing is an essential part of the life cycle of the
software, thus it strongly affects reliability. The majority of tests needs to be
automated, especially endurance and time tests. If the test environment is faulty
or incomplete, it evidently influences the results [5, 6].
• Choice of hardware environment: Software does not exist without hardware. If
the hardware provides an adequate platform for the execution of tasks with
sufficient reserves, the software can work well. There have been cases when the
software configuration worked well in the chosen simulation environment and in
real life for years, but then the hardware ran out of its reserves, and the system
stopped, or worked improperly [5, 6].

1.4 Indices of Software Reliability

Factors influencing software reliability need to be quantified. However, whereas

choosing these factors is straightforward and easy in case of other engineering
fields, finding objective influencing factors is uncertain in this field [7].
Influencing factors include, but are not limited to:
• complexity of the software: It seems easy, but it is not. Indices can be various:
– LOC (Line of Code) and KLOC (Kilo Line Of Code). This data is good as a
first approximation, but it is not definitive. For example the complexity of a
relatively long program with one thread does not match the complexity of
another one with multiple threads.
– Number of the files of the software: Experienced programmers take the
system apart into processes to avoid great and unnecessary complexity and to
make the testing of parts possible.
– Number of tasks running parallel: it is a useful index, but it is not enough in
itself, either.
– Number of IPC elements: These, together with LOC provide a good
approximation, but still not a complete description of complexity. The
number of IPC elements shows the probability of the occurrence of different
special faults (Dead Lock).
• Experience of programmers. This is an unquantifiable index. Great parts of the
software are not written by one programmer, not even one group. Thus, this
important index cannot be estimated.
312 G. Schuster et al.

• Indices of the project management. It is a well-known fact a well-managed

project is likely to have better results. If project management is supportive and
evaluate rationally, development work is done in a more relaxed and effective
way than a work where programmers are asked to report too frequently and in
unproductive ways. The problem is the same here, too: there is no index that can
be given [8].
• Quality aspects: Quality assurance methods and standards can serve as a base for
quality, as well as the CMM certification of the development company.
• The extent of the error: This is the most important index, since it provides an
exact number that shows reliability. Reliability depends on time. The following
types of tests can be applied:
– Stress endurance test. The software is tested in an environment that influ-
ences its operation heavily, e.g. it reduces the resources.
– Load testing and stability test. The software is tested over a long period of
time under peak load conditions.
– Reliability test. It is similar to the previous one, but testing is done under
normal load conditions.
– Regression test. It examines whether the software remains stable if new
elements appear in the system, or if some part of the system starts to work
improperly.
The most problematic feature of these tests is that they take a long time, as they
are statistical procedures. Results are expressed in the following units of mea-
surement: [5, 6].
• Probability of Failure on Demand (POFOD). The probability that the system
will fail when a service request is made.
• Rate of Change of Frequency (ROCOF). It is given for a certain time period.
• Mean Time Between Failures/Mean Time To Failure (MTBF or MTTF).
• Availability: how long a system is available in a given time period and how long
maintenance lasts.
Another problem is the adequateness of the test environment. Obviously testing
cannot be done by humans in the long run, it has to be automated. If the test
environment is incomplete, some parts of the software are not tested, or the auto-
mated tester does not notice differences.

1.5 Faults and Failures

As the definition shows, failures do not always cause faults. It is possible that an
expert notices that the operation of the system operates is slightly different from the
specifications, but it is not certain that the failure manifests itself in the operational
properties of the system. Failures that are apparent on a system level are faults.
Software Reliability of Complex Systems Focus … 313

Faults can be: [9]

• insignificant (unimportant),
• significant (critical),
• known,
• unknown.
Insignificant faults—known or unknown—do not cause serious functional dis-
orders. If, however, the fault is detected, it needs to be corrected soon. In systems
that are not safety-critical faults are sometimes not corrected after a risk assessment.
Minor colouring faults are a good example.
The generator problem of a Boeing 787 is somewhere between failure and fault.
It is a very serious problem, but to have this problem the generator needs to operate
for an improbable length of time. This fault needs to be repaired, but the 787 fleet
does not have to be called back immediately [2, 3].
Critical faults need to be corrected without delay. Unknown critical faults are the
most dangerous. Both the circumstances and their cause are unknown (see the
catastrophe in Seville).
Though prompt correction of faults seems apparent, correction itself carries the
possibility of faults. Validation of the corrected software is essential [9].
It is not apparent why the detection of faults is difficult. The reasons can be:
• the fault only occurs in the joint occurrence of certain circumstances,
• the fault does not occur in normal circumstances/operation, or its effect remains
hidden,
• the fault develops continuously until at some point it causes a system-level
problem.
The first case is typical for concurrent systems. In case of imperfect development
the whole system or parts of it can become blocked in a given constellation.
In the second case the system to which it is not prepared and it may even give an
answer to it. An unfortunate situation like this can be e.g. a dead code.
The third fault can be very banal. The system wastes some resource continuously
and unnoticed. A typical example for this is which leads to the full consumption of
the memory, another example is a memory leak.

2 Increasing Reliability

Increasing reliability is a primary goal. Possible methods are the following:

• The use of appropriate development methods. Observing security and coding
procedures, e.g. MISRA, and complying with the relating standards (e.g. IEC
61508) [9].
314 G. Schuster et al.

• Testing the produced data and programs most exhaustively, even at the expense
of doing more testing projects at the same time. It is even more important in case
of long-term, stress and regression tests.
• The code should be, it should be able to correct small faults so that they do not
lead to serious failures [5, 6].
• The use of software units that have already been used many times and proved
perfect. This is not an ultimate solution, either, because these units can get into
circumstances in which they do not work well.
• The code should contain which can help shed light on more serious faults, and
help prevent the emergence of these faults.
The problem is still persistent, since all redundancies increase complexity, which
in turn is a source of further faults. There have been examples for a security
software rendering a good system unserviceable.

3 Reliability Calculation

Reliability can be practically expressed by the probability of failure, since proba-

bility of well-functioning PW (1) is the complement of probability of failure Pf.
Thus:

Pw ¼ 1  Pf : ð1Þ

In many cases the mathematical analysis of a problem is easier this way.

The first example is a one-thread program, which does not occur in reality, but it
will be of significance in the following examples.
Let us suppose that the statistical data of the program examined are known. In
this case the easiest mathematical approximation of failure (2) is Poisson
distribution:

kk k
P f ðX ¼ k Þ ¼ e ; ð2Þ
k!

k > 0, where is the expected value of failure for the given time period, and is the
k = 1, 2, 3,… number of failures.
Problem: the probability of failure can be computed in this case. Another
question is how we obtain reliable information on the parameter.
Explanation: software packages are well designed and implemented construc-
tions. Thus, conventional observation techniques do not yield useful results. In
these cases people can be examined on the basis of their active program. If there is a
software described above, already in use, with tens of thousands of them running,
conclusions can be drawn concerning parameter k examining collected failure data.
Software Reliability of Complex Systems Focus … 315

Example: let us suppose that 1000 copies of a software are running. The mean
failure rate is one failure every 100 h. In case of one copy, it means one failure for
every 100000 h.
However, it would be preferable for software designers if this value could be
determined before the release of the software, in the design phase.
Question: Can this method be applied to concurrent systems?
The answer is yes, in case of individual program sections, but only to the given
tasks and threads. This problem is further complicated by Inter Process
Communication (IPC).
If we suppose that the failure of an individual software section leads to the faulty
operation of the whole system, the probability of failure (3) is:

X
n
Pf R ¼ Pf : ð3Þ
i¼1

Let us suppose a program with four threads (IPC excluded), all the four threads
of which have the same probability of failure, and a failure leads to the faulty
operation of the whole system.
In this case the probability of failure increases fourfold.1
In case of IPC the relationship between processes constitutes another possibility
of failure. A common occurrence is deadlock between two processes. It can happen
with semaphores and MUTEXes.

3.1 Example

There are two processes: A and B and two MUTEXes: X and y. The priority of B is
higher than that of A. The problem:
• The process is running, and gets MUTEX X.
• The scheduler stops process A, because process B is ready, and starts B.
• Process B gets MUTEX Y.
• Process B is blocked on MUTEX X.
• Process A is running, since B is in blocked state.
• Process A is blocked on Y MUTEX
Process A cannot run, because it is waiting for the release of MUTEX Y, and for
this reason it cannot release MUTEX X, on which process B is waiting, and so B
cannot release MUTEX Y.

1
Increase in complexity leads to decrease in reliability.
316 G. Schuster et al.

This is the simplest example; it is easy to notice this problem and handle it.
A more difficult problem is when this phenomenon does not occur in pairs, but
more processes get into the cyclic blocking state.
The probability of this state can be estimated. The following method is suitable
for two processes: Let us suppose that both A and B processes use X and Y
MUTEXes with normal distribution. Based on that:
• In case of normal running, the probability of process A being in a field protected
by MUTEX X is PX .
• In case of normal running, the probability of process B being in a field protected
by MUTEX Y is PY .
• The probability of running of process A is PA .
• The probability of running of process B is PB .
• The probability of processes A and B running at the same time is

PAB ¼ PA PB : ð4Þ

• The probability of deadlock (5) of the two MUTEXes is

Pdeadlock ¼ PA PB PX PY : ð5Þ

The solution does not seem complicated, but the result is unfavourable from the
point of view of running reliability. From the point of view of testing, it is
advantageous, since the error is detected quickly.
The above example is perfunctory. With tens or hundreds of processes the
probability of an error like this decreases significantly, but so does the chance of its
detection, too.
Results are similar, if the hardware is taken into consideration as well.
Processing a state originating from the hardware takes time, too.
A typical case is the processing of network allocation, when a device finds the
network device available, but by the time it would start to use it, another device has
already started to use it. This time interval is called a soft time slot.
Unfortunately, this problem also happens when this task is performed by
hardware. It can be solved by different methods, like CSMA/CD or CSMA/CA.
Depending on the usage of the network device, intolerable delays may occur in firm
and hard real time cases.
It happens when the resources of the system are not sufficient in the critical time
period. If the hardware and software resources suffice, the system can handle the
most disadvantageous state, and does not experience time related disturbances. Our
method of computing is the following:
Software Reliability of Complex Systems Focus … 317

qi is the relative frequency of resource usage (6), where

P
j sij
qi ¼ ; ð6Þ
Ti

sij is the time span of the critical phase, during which the problem may occur
(regarding the ith process). If another process wants to access the resource, it
blocks access
Ti the time span examined (regarding the ith process)
The limit at Ti infinity of pi is the probability of the occurrence of the critical
phase.
There is an error or deviation, when a certain number of processes clash.
If pi probabilities are the same for the access part resource, the binomial distri-
bution (7).
!
n k
pik ¼ p ð1  pi Þnk : ð7Þ
k i

For the error or deviation to happen, k processes out of n must happen during the
given time span. However, there is an error, if more than k processes enter at the
given time.
!
X n
n
pf ¼ pij ð1  pi Þnj ; ð8Þ
j¼k j

where pf is the probability of the error or deviation (8).

To avoid it, sufficient resources need to be allocated during the designing of the
system, so that no problems can arise even with maximum load.
Memory leaking is the next case to be examined. The cause of the deviation—
and later the error—is that the process fails to release memory it has been allocated,
thus the available memory is diminishing.
The error model is a simple subprogram, which has many exit points. This
subprogram exits at one of the exit points—where the memory is not released—
with pi probability, thus reducing available memory, pi can also be estimated using
relative frequency, supposing uniform distribution.
Let us suppose that:
Ma is the memory available,
Δm is the allocated, but not released memory,
k the number of runnings,
s the average “cycle” time of runnings,
Tf the expected interval of the error
318 G. Schuster et al.

The available memory must come to an end for the error or deviation to occur.

Ma
0 [ Ma  kDD; where Tf [ s : ð9Þ
Dm

Memory leaking is an error. However, software developers often do not solve

the problem, but advise the restart of the system from time to time.
The next issue is the examination of software as a Markov process.
Markov process is a stochastic process, the future behaviour of which depends
only on the present state of the system [10]. In short, it is described as “memory
less”.
This boundary condition is not met by the software. A simple example for this is
the resource usage of the system for one moment of time.
Let us look at the following scenario: there is a process that has considerable
memory access. The data storage device becomes fragmented due to high load, this
way its access time gets longer. It is, without doubt, an effect from the past.
If the system has layered structure, the “upper” layers have supervisory and
correcting functions—especially in case of SIL3 and SIL4 [11]. It is not suitable to
treat the system as a Markov process in this case, either.
These were all examples of problems which lead to deviations or failures due of
some designing error. In all cases the probability of the occurrence of the sub
process was attempted to be established. This probability value is, however, more
difficult to determine, since the probability of the occurrence of these critical phases
is so small that it would take years to observe the given process, or large number of
elements need to be monitored in order to determine the desired value.
A further problem is that software packages are without an exception concurrent
system. Thus, observing them is a far more complex task than observing any other
traditional technical systems.
Software manufacturers and users would like to determine risk elements before
the release of the software. The methods described above are not suitable for this
task.
The solution is the monitoring of the producer of the software. The producer
most certainly has many software products, several of which run under real con-
ditions. Collecting these data allows for a statistical sample that can be analysed.
Our study aims to examine how arbitrary software errors are, and what kind of
trends can be observed considering the software manufacturer.
If we consider the software development procedure as a time series, we can
apply Hurst analysis [12].
Note: Hurst analysis has been designed for time series, but since software
development is not time dependent enough to examine time spans, time is substi-
tuted by the number of code lines.
The number of occurrences of errors and deviations has been determined
regarding a given number of code lines.
Software Reliability of Complex Systems Focus … 319

Ei the number of errors and deviations regarding a given number of code lines,
i ¼ 1; 2; . . .; m
m the number of time series,
N the number of code lines of samples,
EN average number of errors (10)

1X m
EN ¼ Ei : ð10Þ
m i¼1

Let us calculate the cumulated averages of the given periods (11).

X
m
X ði; N Þ ¼ Ei  EN : ð11Þ
i¼1

Then let us calculate the difference of the maximum and minimum value of
X ði; N Þ.

R ¼ max X ði; N Þ  min X ði; N Þ: ð12Þ

This value is normalised with the standard deviation of the whole series:

R
¼ NH ð13Þ
SN

H is the Hurst exponent, the value of which is characteristic of the errors and
trends.
H = 0.5 indicates a completely uncorrelated series; that errors occur arbitrarily.
H > 0.5 indicates a time series with long-term positive autocorrelation, i.e. if
errors tend to decrease in a given time period, the same is expected in the next
period of time.
H < 0.5 predicts a switching between high and low values in adjacent pairs. That
is, a single high value will probably be followed by a low value, and that the value
after that will tend to be high.
This is method is promising. However, it has two problems. The producers of the
software tend to keep smaller errors for themselves, since this way their reputation
will not be damaged. The other problem is that users may not become aware of
deviations, or may not deal with the problem. In both cases the statistical sample
becomes less accurate.
320 G. Schuster et al.

4 Consequences and Conclusion

Software reliability can be expressed with the help of probability attribute. It is

difficult to find a model that enables us to estimate the value fairly. Having
examined the question from a multitude of viewpoints it can be stated that there is
no one perfect theory. It is difficult to find a model that enables us to estimate the
value fairly. Having examined the question from a multitude of viewpoints it can be
stated that there is no one perfect theory.
The application of the mathematical solutions explained in this article requires
some preliminary knowledge, which cannot be obtained with 100% certainty.
The method that we were studying aims at the producer of the software, since the
statistical examination of its product yields results that can form the basis of further
research on the software.
Reliability can be increased with different methods, rules and testing, but
experience shows that just like other man-made systems, reliability of software
packages cannot be 100%. Reliability is usually unchanging after the phase of
design, but the fact that codes depend on humans is a serious problem in itself.
The reliability of a software is probability. It is difficult to find a probability
model that gives a good estimate of the software’s reliability. Having examined this
topic, we can assert that it is impossible to name one perfect theory. Different
methods, rules and testing can increase reliability, but experience shows that—
similar to other artefacts made by humans—software packages cannot be 100%
reliable.

References

1. Flottau J, Osborne T (2015) Software cut off fuel supply in stricken A400M. http://
aviationweek.com/defense/software-cut-fuel-supply-stricken-a400m. Accessed 19 May 2015
2. The Aviation Safety Network, Accident description. http://aviation-safety.net/database/record.
php?id=20150509-0. Accessed 19 May 2015
3. Mouawad J, FAA (2015) Orders fix for possible power loss in boeing 787. http://www.
nytimes.com/2015/05/01/business/faa-orders-fix-for-possible-power-loss-in-boeing-787.html.
Accessed 19 May 2015
4. Lions JL (1996) ARIANE 5 Flight 501 Failure. https://www.ima.umn.edu/*arnold/disasters/
ariane5rep.html. Accessed 19 July 2015
5. Pan J (1999) Software reliability. https://users.ece.cmu.edu/*koopman/des_s99/sw_
reliability/
6. Schneidewind FN (1997) Reliability modeling for safety critical software. IEEE Trans Reliab
46(1)
7. Van Solingen R, Berghout E (1999) The goal/question/metric method: a practical guide for
quality improvement and software development. McGraw-Hill International
8. Long J (ed) (2008) Metrics data program, national aeronautics and space administration.
http://mdp.ivv.nasa.gov/index.htm
Software Reliability of Complex Systems Focus … 321

9. Tokody D, Schuster G, Papp J (2015) Study of how to implement an intelligent railway

system in Hungary. In: 2015 IEEE 13th international symposium on intelligent systems and
informatics (SISY), Subotica, pp 199–204. doi:10.1109/SISY.2015.7325379
10. Markov process (mathematics)—Britannica Online Encyclopedia
11. Marszal EM, Scharpf EW (2002) Safety integrity level selection—systematic methods
including layer of protection analysis. The Instrumentation, Systems, and Automation
Society, Research Triangle Park, NC, USA. ISBN: 9781556177774
12. Szilágyi GyA (2016) The fractal dimension of the air accidents. Repüléstudományi
Közlemények XXVIII(2):41–48. HU ISSN 1417-0604
Usage of an Optical Flow Sensor
in Robotics to Define Orientation

Máté Koba, Roland Bartók and László Czap

Abstract In autonomous robotic control it is necessary to measure the exact tracks

of the robot and then calculate the best route for it. This article is about to introduce
a method which can be used to measure the route of an Unmanned Ground Vehicle
to help the controller software to guide the vehicle. For now, the most popular way
to measure and track of a vehicle or a robot’s movement is the incremental or
absolute encoder installed on the vehicle’s wheels. This method measures move-
ment in an indirect way because there can be gear system on its motor or in
omnidirectional systems, there can be a slip between the wheel and the ground so
the data of these sensors could be easily tampered. With this method we install an
optical flow sensor (or sensors) to the undercarriage of the robot. This way the
system measures the movement in a fully direct way because the sensors detect the
movement between the robot and the ground and there is no mechanical or elec-
trical connection between the sensor and the reference.

1 Introduction

This article explains the usage of an optical flow sensor through an example of a
custom-built robot. The robot built for a Hungarian robot-building contest called
“Magyar Alkalmazott Mérnöki Tudományok Versenye”. The software of the robot
is running ROS (Robot Operating System), it is controlled over wireless commu-
nication and it has autonomous functions too. In the following, there will be written
about the robot’s structure and the optical flow sensor based navigation system will
be explained thoroughly.

M. Koba (&)
R. Bartók
L. Czap

Institute of Automation and Infocommunication, University of Miskolc,
Miskolc, Hungary
e-mail: autkmate@uni-miskolc.hu
R. Bartók
e-mail: bartokroland@hotmail.hu
L. Czap
e-mail: czap@mazsola.iit.uni-miskolc.hu

© Springer International Publishing AG 2017 323

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_29
324 M. Koba et al.

2 Introducing the Robot

The task of the robot is manoeuvring on a track which has 15 pillars made of PVC
and it has to occupy as many pillars as it can by putting a beacon into them. The
robot is equipped with 3 omnidirectional wheels (omniwheel) so it is capable of
complex movements. There are 3 main stages of the robot. At the lowest level there
are the 3 motors equipped with omniwheels, 2 optical flow sensors and the
battery-pack, shown in Fig. 1. On the middle level there are the computing units, an
STM32F4 Discovery board and a Raspberry PI 3. The top level contains the beacon
on a mechanism which is responsible for putting the beacon in the right position.

2.1 Why Optical Flow Sensors?

For a precise navigation it is very important to know the exact orientation of the
robot. In robotics there are a lot of methods of defining the position and orientation
of a robot such as using inertial sensors, GPS navigation based applications but
these are quite expensive and easily jamming appliances. Using optical flow sensor
is much cheaper than the above mentioned ones and it is less sensitive to electrical
interference still it is an accurate sensor. With these sensors it is possible to measure
velocity of multiple-driving systems and make measurements of conveyor systems
like in [1].

Fig. 1 Bottom level sketch

without battery and sensors
Usage of an Optical Flow Sensor in Robotics to Define Orientation 325

3 Optical Flow in Theory and Presenting an Actual Sensor

3.1 Optical Flow Phenomenon

Optic flow is a visual event which can be noticed during our everyday life. Optic
flow is the visual motion we can experience during movements. It can be introduced
to a simple situation. Suppose that you are sitting in a car or a train and looking out
the window. The objects outside like trees, buildings, etc., seem to move back-
wards. The motion you experience is optic flow. From this motion, you also could
tell your distance from these objects. Distant objects appear still and closer objects
appear to move backwards, faster than the objects far away from you. There are
mathematical relationships between the magnitude of the optic flow and the chosen
object’s relation from the spectator. When the speed of travel doubles, the optic
flow you see will also double. If an object is twice as close as it was, the optic flow
will double again. The magnitude of optic flow also depends on the angle between
the direction of travel and the direction of the inspected object. Let’s see the case of
travelling forward on Fig. 2 [2].
The optic flow is the largest when the object is to the inspector’s side by 90° or
directly above or below from it. In front of the inspector the optic flow will be zero.
So the object in front of the inspector will appear to be still.

Fig. 2 Optical flow in case

of yaw right and moving
forward [2]
326 M. Koba et al.

3.2 Operation of an Optical Flow Sensor

Optical flow sensors could find in optical mice. These sensors could be called
intelligent sensors, they process the texture of the surface the sensor is inspecting
and after that they define the relative movement of the sensor and the surface at the
x, y plane. The camera makes pictures sequentially and compares the data in the
order to the sequence. According to the texture and the time between the two
pictures it is possible to calculate the amount and the direction of the movement the
object travelled. With further calculations, we could define current speed, object
orientation from start point, turning angle [3, 4].

3.3 Benefits of Using These Sensors

With flow sensors, the calculations of speed and movement are much more punctual
than using encoders on the motors or using GPS even on hard terrain. In our case,
with the omnidirectional wheels to measure speed and distance is more simple and
accurate using optical flow sensors rather than encoders. With omniwheels the
direction of travel is not clearly definable from the rotation of the motors because in
this case the robot is not heading to the exact direction the wheels are spinning.
Also during each round of the competition there are four robots on track and it is
possible that robots collide or push each other. When the robot is pushed or there is
any slip between the wheel and the track the encoders provide wrong data for the
speed and orientation calculations. It is also possible to use GPS but commercial
GPS provides weak signal inside buildings. It is also possible to use sensor-fusion
to define movement, but it requires a lot of different types of sensors which are very
sensitive and easily jammable. The best way to measure speed and travelled dis-
tance is to use optical flow sensors [5].

3.4 ADNS-3080

The sensor is based on Optical Navigation Technology, which measures changes in

position by optically acquiring sequential surface images (frames) and mathemat-
ically determining the direction and magnitude of movement. It contains an Image
Acquisition System (IAS), a Digital Signal Processor (DSP), and a four-wire serial
port. IAS acquires surface images via a lens and illumination system. The images
are processed by the DSP to determine the direction and distance of motion.
The DSP calculates Dx and dy relative displacement values. With an external
microcontroller it is possible to read the Dx and dy information via the sensor serial
port [6]. The ADNS-3080 block diagram is shown in Fig. 3.
Usage of an Optical Flow Sensor in Robotics to Define Orientation 327

Fig. 3 Block diagram [6]

The sensor has 5 main parts:

• serial port
• ctrl logic
• image processor
• power circuit
• oscillator
The sensor structure built up by 3 parts:
• Image Acquisition System (IAS)
• Digital Signal Processor (DSP)
• 4 wire serial port (SPI)
After the reflected light from the illuminated surface reaches the sensor the IAS
acquires the images and then the DSP processes the image and the processor
calculates the magnitude and direction of the movement and after that it is possible
to calculate the relative movement [6].

3.5 Sensor Sensitivity

The sensor is not equally sensitive at the entire light spectrum. The figure below
shows the sensor sensitivity curve related to light wavelength. The sensor is most
sensitive in the 600–700 nm range. The sensor sensitivity is shown in Fig. 4 [6].
328 M. Koba et al.

Fig. 4 Sensor sensitivity [6]

4 Using the Sensor

4.1 Sensor and Lens

The sensor is accessible with a PCB mount so the main circuits and the camera are
attached and the user could change lenses and connect the sensor to a micropro-
cessor. The mounted sensor is shown in Fig. 5.
There are a wide variety of lenses compatible with this mount. The lens image
shown in Fig. 6 at the farthest and nearest focal point, the grid is 1 mm  1 mm.

4.2 Guarantee Appropriate Light Conditions for the Sensor

Since the sensor is very sensitive to the light conditions it is necessary to provide
eligible amount of light for it. The light is provided by 4 LEDs each has 2180–
4200 mcd intensity and 60° radiating angle and their wavelength is 619–629 nm.
Every 4 pieces are mounted in a 3D printed case which provides the optimal angle
for each LED. It is necessary to notice that every lens needs a different placement
according to its optical specification. The correct LED angle is calculated by the 3D
designing program. Figure 7 shows the 3D model of the LED case.
Usage of an Optical Flow Sensor in Robotics to Define Orientation 329

Fig. 5 Sensor PCB

Fig. 6 Selected lens image over test grid

4.3 Sensor Placement

There are 2 sensors on the robot. The sensors are on the x-axis of the robot, rotated
by 90° to each other. The result of the rotation is that the given data is compensated
and this way it is possible to calculate rotating angle besides x and y distance.
Figure 8 shows one of the attached sensors.
330 M. Koba et al.

Fig. 7 LED case

Fig. 8 Sensor placement

Applying lenses with longer focal distance is very good when the quality of the
surface is not good or there is some stain or some sort of roughness on it. With long
focal length the depth of field and the field of view is bigger, the sensor could see
larger area and less sensitive for stains.
Usage of an Optical Flow Sensor in Robotics to Define Orientation 331

5 Mathematics

To define the exact distance, speed and angle values the microcontroller has to
calculate them from just the two sensor’s Dx and Dy data. After placing the sensors,
we could define coordinate systems referring to the sensor and robot orientation
(Fig. 9). Below is shown the references [7].
From the coordinate system we could make equations referring to x (1) and y (2)
movement, the following calculations are from [7].

xi ¼ DXR
sinðhi þ Ui Þ  DYR
cosðhi þ Ui Þ  Dx
ri
cosðhi þ Ui Þ ð1Þ

yi ¼ DXR
cosðhi þ Ui Þ þ DYR
sinðhi þ Ui Þ þ Dx
ri
sinðhi þ Ui Þ ð2Þ

For the computing we should redistribute the equations and make a vector form
for the constants as in (3).
2 3 2 3 2 3 2 3
xi sin ðhi þ /i Þ  cos ðhi þ /i Þ r cos /i
6 yi 7 6 cos ðh þ /Þi 7 6 7 6 7
6 7 ¼ D XR 6 i 7 þ D YR 6 sin ðhi þ /i Þ 7 þ Dx6 r sin /i 7 ð3Þ
4 xi 5 4 sin ðh þ /i Þ 5 4  cos ðhi þ /i Þ 5 4 r cos /i 5
i
yi cosðhi þ /i Þ sin ðhi þ /i Þ r sin /i

After that we separate the constants and the variables (4).

2 3
sinðhi þ /i Þ  cosðhi þ /i Þ r cos /i
6 cosðhi þ /i Þ sinðhi þ /i Þ r sin /i 7
A¼6
4 sinðhi
7 ð4Þ
þ /i Þ  cosðhi þ /i Þ r cos /i 5
cosðhi þ /i Þ sinðhi þ /i Þ r sin /i

For the absolute distances we need to multiply the sensor values with the
Moore-Penrose inverse of matrix A (5).

Fig. 9 Coordinate systems

332 M. Koba et al.

A þ ¼ AT ðAAT Þ1 ð5Þ

The final equation defines the absolute distances and rotating angles from each
iteration when the microcontroller’s program measures the movement (6).
2 3
2 3 x1
DX R 6 y1 7
4 DY R 5 ¼ A 6 7
þ
ð6Þ
4 x2 5
Dx y 2

6 Conclusion

The system we designed is working well, the provided data is good for measuring
movement. The provided data is more reliable than the encoder applications. The
system is sensitive to ambient light so it is needed to provide proper light conditions
for the sensor, especially in dark environment. For best operation, using light
sources with the right wavelength and measure and set focal length precisely is
necessary. For more precise operation, more sensors in different orientation could
be used.

Acknowledgements This project has received funding from the European Union’s Horizon 2020
research and innovation program under grant agreement No 691942. This research was partially
carried out in the framework of the Center of Excellence of Mechatronics and Logistics at the
University of Miskolc.

References

1. Németh J, Illés B (2015) Determination of the ratio of centripetal forces in the friction drive
used at more places. XXIX. In: microCAD international multidisciplinary scientific conference,
University of Miskolc, Miskolc (in Hungarian)
2. Centeye (2013) [Online] http://www.centeye.com/technology/optical-flow/
3. Sorensen DK (2004) Texas A&M University, on-line optical flow feedback for mobile robot
localization/navigation
4. Sekimori D, Miyazaki F (2007) Precise dead-reckoning for mobile robots using multiple
optical mouse sensors. Informatics in control, automation and robotics II. Springer, pp 145–151
5. Tresanchez M, Pallejà T, Teixidó M, Palacín J (2009) The optical mouse sensor as an
incremental rotary encoder. Sens Actuators A 155(1):73–81
6. Avago, ADNS-3080 Datasheet (2007)
7. Bell S (2011) High-precision robot odometry using an array of optical mice
Pose Determination for Autonomous
Vehicle Control

Ahmed Bouzid, József Vásárhelyi, Roland Bartók and László Czap

Abstract For the purpose of determining the position and orientation of a moving
robot and autonomous vehicle, inertial sensors and magnetometer data are com-
puted in order to enhance GNSS (Global Navigation Satellite System) data accu-
racy. This paper presents a method called hybrid localization that combines
absolute localization, using exteroceptive data, and dead reckoning technique, using
proprioceptive data. A positioning method based on dead reckoning technique is
developed in this paper.

1 Introduction

Nowadays, objects in movement (autonomous robots in particular) have to know in

real time their navigation parameters for better trajectory control. The parameters of
navigation intended are: orientation, position, velocity and acceleration. The pro-
cess of calculating these parameters is called pose determination. Autonomous
vehicles should consider also the dragging force, which acts in the drivetrain of the
system [1].
In the case of the navigation, IMU (Inertial Measurement Unit) is used for data
acquisition to estimate pose information (all those parameters above mentioned
parameters) of a moving object [2]. The first application of an inertial unit was a
directing system for a torpedo. The system is based on a gyroscope patented by
Ludovico Obry in 1894. The usage of inertial units was limited due to high prices

A. Bouzid (&)
J. Vásárhelyi
R. Bartók
L. Czap

University of Miskolc, Miskolc, Hungary
e-mail: qgebouzid@uni-miskolc.hu
J. Vásárhelyi
e-mail: vajo@uni-miskolc.hu
R. Bartók
e-mail: qgeroli5@uni-miskolc.hu
L. Czap
e-mail: czap@uni-miskolc.hu

© Springer International Publishing AG 2017 333

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_30
334 A. Bouzid et al.

until MEMS (Micro-Electromechanical Systems) technology appeared. MEMS

devices are very tiny and inexpensive, but they provide some systematic and ran-
dom noises and generate many errors. To solve this problem we need some tools
and implement adequate software/hardware architecture.
For a mobile robot or an autonomous vehicle the collision avoidance is very
important. A wide range of sensors are used for pose determination, distance to
objects, walls and collision avoidance. Better results are obtained from different
type of sensors (for example infrared, ultrasonic and laser distance sensors). Using
many sensors allows sensor fusion to obtain higher precision results.
After pose determination, the robot behaviour control is relatively easy by using
the FRI based Behaviour Description Language or the Bayes-classifier [3].
This paper is organized as follows: the first section gives details about the
proposed method for the pose determination and control. Then theoretical back-
ground is presented, which is followed by the presentation of the simulation results.
Finally, some conclusions are presented. In the paper the expressions autonomous
vehicle and autonomous robot are considered equivalent expressions.

2 Proposed Method for Pose Determination

Pose determination is a complex process; it is divided into 4 subsystems. It requires

and IMU (Inertial Measurement Unit) of 9 sensors (3 accelerometers, 3 gyroscopes
and 3 magnetometers) and GNSS (Global Navigation Satellite System) receiver. In
order to process the navigation parameters, the processing unit needs hybridization
of sensors data to get accurate estimation. The subsystems are (see Fig. 1) [4]:

Fig. 1 Block diagram of the pose estimation solution

Pose Determination for Autonomous Vehicle Control 335

• GNSS processing (A): uses a GNSS receiver to process the position and the
velocity;
• GDOF processing (B): the Gradient-Descent Orientation Filter is an algorithm to
estimate the orientation using the 9 sensors above-mentioned [5].
• Mechanization (C): is the step intended to estimate the position and the orien-
tation using 3 accelerometers and 3 gyroscopes.
• EKF processing (D): Extended Kalman Filtering is a robust algorithm that
estimates the states of the system in a noisy environment.
In Fig. 1 the following notations were used: PGNSS ; VGNSS are respectively the
position and the velocity after GNSS processing; PINS ; VINS are respectively the
position and velocity given by the mechanisation; dp and dv are the difference
between the results of GNSS processing and Mechanisation; x, a are respectively
the angular velocity and the acceleration given by the 6 DoF IMU; m is the data of
the magnetometer. W is the orientation vector. Δp, Δv and Δ W are the errors
computed by the EKF.
In the paper 2D positioning is treated in the plan map of the autonomous vehicle.
In real time, the robot calculates his position and stores it in order to reconstitute its
trajectory. The block diagram of the proposed system is given in Fig. 2. The
processing unit uses raw data of the accelerometer sensors to measure the accel-
eration of the robot. The used sensor is ADXL335 that is a set of 3 accelerometers
(one for each degree of freedom) orthogonally placed on a single chip providing
analog data. A secondary processing unit (ZYNQ device) is responsible for the
acquisition of the analog data (analog to digital conversion) from sensors and the
position estimation. Then the processing unit puts the processed results in a serial
frame and sends via the UART to the main processing unit (STM32). For more
accurate data, the main processor sends an interrupt to the secondary processing
unit each time the wheels stop to reinitialize the dead reckoning equations and avoid
divergence.
The primary processing the STM32—ARM Cortex M4 based board, while the
secondary boar is a Zynq System on Chip contain a dual ARM Cortex A9 MP Core
and Neon Single precision Floating point unit for each processor unit. The Zynq
also contains Programmable Logic (PL) for user based algorithm implementation.
The Zynq based board is Zedboard, which finally will do all the computation for the
robot.

Fig. 2 Block diagram of the system

336 A. Bouzid et al.

3 Mathematical Background of the Dead Reckoning

System

Knowing that the calculations are iterative, the equations for dead reckoning are:

x ¼ 12 ax dt2 þ v 
x dt þ x . . .
ð1; 2Þ
y ¼ 2 ay dt þ vy dt þ y . . .
1 2 

where x, y are the coordinates (position) of the material point (the Robot); ax, ay are
the accelerations measured by the sensors, dt is the sampling time; v 
x ; vy are the
 
velocities calculated during the previous iteration, x ; y finally are the coordinates
of the previous position (Eqs. (3) and (4)). The velocities from Eqs. (3) and (4)
must be injected in the Eqs. (1) and (2) on each iteration. The velocities are
measured as follows:

vx ¼ ax dt þ v
x ...
ð3; 4Þ
vy ¼ ay dt þ v
y ...

4 Simulations

In order to validate the algorithm, the system has been modelled in Matlab Simulink
and simulated considering ideal conditions (non-noisy environment). The robot
moves on a 2D plane (X and Y axis) and starting from the origin (0, 0).
In the simulation process, some conditions have been supposed and they were
considered three scenarios (with the sampling rate of dt = 0.1 s during 100
iterations):

4.1 First Scenario

Figure 3 describes the trajectory of the robot with the following conditions: initial
velocity for X axis is vxo ¼ 1 m/s and the acceleration on the y axis is ay ¼
9:8 m s2 (with vyo ¼ 0; ax ¼ 0). The simulation results are given below with the
already mentioned scenarios.
The graphic looks like the trajectory of a horizontally launched projectile.
Pose Determination for Autonomous Vehicle Control 337

Fig. 3 Trajectory of the robot on the 1st scenario

Fig. 4 Trajectory 2 of the robot on the 2nd scenario

4.2 Second Scenario

The simulation is done without initial velocities. The variations on the accelerations
are given in Fig. 6 (left part).
(2nd Scenario: vxo ¼ vyo ¼ 0; ax1 ¼ 1 m s2 ; ax2 ¼ 1; ay1 ¼ 1; ay2 ¼ 1)
Figures 4 and 5 describes the trajectory of the robot with the conditions
described in Fig. 6 without initial velocities vxo ¼ vyo ¼ 0:
338 A. Bouzid et al.

Fig. 5 Trajectory of the robot on the 3rd scenario

Fig. 6 Accelerations versus number of iterations (Left 2nd scenario. Right 3rd scenario)

4.3 Third Scenario

The simulation is done without initial velocities. The variations on the accelerations
are given in Fig. 6 (right part).
(3rd Scenario: vxo ¼ vyo ¼ 0; ax1 ¼ 1 m s2 ; ax2 ¼ 20; ay1 ¼ 1; ay2 ¼ 5)
Pose Determination for Autonomous Vehicle Control 339

5 Conclusions

The simulation results confirmed our expectations. The dead reckoning system
algorithm gives good results in ideal environment. There were presented three
scenarios for robot movement in three different conditions that were studied for the
evolution of acceleration versus the number of iterations.
The next step will be to model the accelerometers and introduce noise in the
system to be close to real conditions. Then implement the solution in ZYNQ device
in order to experimentally validate the proposed model.

Acknowledgements The research work was (partially) supported by the Hungarian Scientific
Research Found grants OTKA 29326 and Fund for the Development of Higher Education FKFP
8/2000 project. This research was (partially) carried out in the framework of the Center of
Excellence of Mechatronics and Logistics at the University of Miskolc.

References

1. Németh J, Illés B (2015) Determination of the ratio of centripetal forces in the friction drive
used at more places. XXIX. In: microCAD international multidisciplinary scientific conference,
Miskolc, University of Miskolc (in Hungarian)
2. Noureldin A et al (2009) Performance enhancement of MEMS-based INS/GPS integration for
low-cost navigation applications. IEEE Trans Veh Technol 58(53):1077–1096
3. Bartók R et al (2016) Embedded behavioral model implementation. In: Proceedings of the 17th
international carpathian control conference (ICCC), Slovak Republic. IEEE
4. Bouzid A et al (2016) Implementation of INS/MAG/GNSS hybridisation technique for pose
determination based on SoC and low cost sensors: theoretical approach and synthesis. In: 17th
Carpathian control conference (ICCC), Slovak Republic. IEEE
5. Madgwick SOH (2010) An efficient orientation filter for inertial and inertial/magnetic sensor
arrays. University of Bristol, UK
Description of a Method for the Handling
of Customer Needs in Logistics

Béla Illés, Róbert Skapinyecz and György Wagner

Abstract The paper describes the application of the QFD method, a technique used
for the evaluation and proper realization of the different customer expectations, in
the quality management of logistics systems. Both the theoretical basics of the
method, as well as the main steps of its implementation are introduced. The
implementation itself is presented with the help of a practical example that is
strongly related to both the logistics and the automotive industries, as the latter
especially relies on complex supply chains that require the extensive utilization of
quality management tools. Besides the previous, the paper also provides an over-
view of all the possible areas of utilization for the QFD in the logistics industry.
Therefore, the described method can have a great value from both the academic and
the industrial perspectives.

1 Introduction

By the present day, logistics has become a common concept. The word “logistics”
can be seen on notices on the vehicles of transport companies. The excellent
“logistical quality” is advertised on the stationery posters of facilities. But really,
what is in the background of the concept of “logistics”?
Logistical problems and tasks arise in all areas of the industry and economics.
This puts a lot of requirements on the comprehensive field that is today called as
logistics. In the meantime, logistics is under constant change due to the techni-
cal innovations and the changing social and political boundary conditions.

B. Illés (&)
R. Skapinyecz
G. Wagner

University of Miskolc, Miskolc, Hungary
e-mail: altilles@uni-miskolc.hu
R. Skapinyecz
e-mail: altskapi@uni-miskolc.hu
G. Wagner
e-mail: wagner@iit.uni-miskolc.hu

© Springer International Publishing AG 2017 341

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_31
342 B. Illés et al.

The diverseness and the dynamics present a great challenge for the logistics pro-
fessionals of the future, as intelligent and economical solutions are awaited from
them.
Logistics assures,
• the flow of materials and the connected information,
• the flow of waste materials and the connected information,
• the change of the location of objects (persons, animals, things) in order to avoid
losses (quantitative differences), to avoid changes in the quality of objects
(damage), while in the meantime new objects are not created. Logistics can be
distinguished from manufacturing or processing procedures which are generally
aimed at the production or modification of new products.
As it can be seen from the previous, the quality of logistics processes affects the
overall performance of a logistics system in multiple ways, therefore it is an integral
part of the value creation process in logistics [1]. Quality itself can be basically
measured through customer expectations. For these reasons, the proper handling of
the customers is essential for all companies. In the long term, the company will only
be successful if it perfectly knows its customers and their expectations, and it
shapes its products, processes and systems in such a way that they properly and
effectively fulfil these expectations.
Many companies already applied special software for the registration and
maintenance of their client data. The purpose of using such systems is to attain the
expected benefits and income, thereby assuring the success of the company. The
variability of such software is very large. Its capabilities span from the simple
customer data base to data mining tools which generate new implications from the
customer data, while also including work processes which realize automatic
information distribution. These techniques became especially important today with
the rapid deployment of big-data applications.
One of the tools that can utilize the results of the previous techniques is the QFD
(Quality Function Deployment), a method that is also known in the automotive
industry. It basically provides a controllable way to effectively transform the
expectations of the customers into measurable technical data. QFD relates to
the first principle of Lean thinking, which is identifying the customer and the
customer’s value [2, 3]. In the paper, the most important theoretical basics of this
method will be described, together with a practical example which is related to both
the logistics and automotive industries.

2 QFD—A Method for the Realization of Customer

Expectations

QFD is an abbreviation standing for Quality Function Deployment. It originally

came from Japan, the year of its appearance was 1966. In Japanese, the term means
Hin Shitsu, Ki No, Ten Kai, which means the “Presentation of Characteristic
Description of a Method for the Handling of Customer Needs … 343

Availability (direct)
Duration of transport Transport capacity
Transport location Transport flexibility
Transport address Level of transport
service
Contact with transporter State of transport
Transparency of
transport Additional performances
Reaction to Price-performance Transparency of
transport failures ratio contracts

Fig. 1 Customer’s expectations as the starting points of QFD

Fig. 2 QFD—basic
philosophy Market / Customer
Needs, expectations, Wishes
requirements

Formulation of the customer

requirements for the product,
What ? performance

Interdisciplinary work-group

How ? Translation of the re-

quirements to the company’s
language

Functions”. The founder of the QFD was Yoji Akao. His book on the QFD was
presented in Japan in 1978 (see [4]). A German edition of this book was also
published, first in 1992 [5].
The goal of the QFD method is to properly select the customer’s expectations
and reformulate them into technical or organizational solutions. Therefore, the
starting point for the QFD are the customer’s expectations themselves (Fig. 1).
However, the knowledge of the customer’s expectations alone is not enough, as
they consistently have to be realized as well.
The QFD method has a systematic, multi-step procedure. Generally, it can be
stated that with the help of the QFD, the question “What should be done?” is
reformulated into the question “How it should be done?” (Fig. 2).
344 B. Illés et al.

Table 1 Advantages and disadvantages of the QFD

Advantages Disadvantages
Assures the general customer Significant time expenditure
orientation at the company
Preventive disposition Personal expenditure
Sector dependent The method is less known outside quality
management
Systematic, proven procedure The results are dependent on the competencies and
the collaboration skills of the participants
Applicable in multiple manner The purpose oriented usage requires the special
knowledge of the method
Interactions are consciously It partially requires significant abstraction skills
investigated
Needs team work and demands It reduces the investigation to a few relevant
communication over different areas parameters

The QFD aims to achieve both the consistency and completeness of the goal. For
this reason the best is to utilize team work with the participation of all the concerned
partners. For instance in case of the development of a new product, the team needs
to develop all of the followings: the marketing, the construction, the work prepa-
ration, the production, the maintenance and the service background. The usage of
the QFD in logistics also covers the marketing, the logistics planning and the
involved logistics areas, among them controlling.
In the literature [4], three starting points are differentiated in case of the QFD:
• the excluded start according to Akao (comprehensive QFD),
• the 4-phase model according to the American Supplier Institute (ASI),
• the King matrices [6].
At this place we do not wish to compare these different approximations. Table 1
lists and summarizes the advantages and the disadvantages of the QFD.
In the followings, the ASI 4-phase model will be described, which is often used
in industrial environments and for that reason it is especially applicable in the field
of logistics as well.

3 General Procedure of the QFD

The individual phases of the ASI 4-phase model, together with their inputs and
results are shown in Table 2.
From Table 2 it can be seen that the most important outputs of the individual
phases are the inputs of the following ones. Thereby the continuous stream of
arrows is realized.
For transparent documentation, the usage of a worksheet has proven to be the
best practical solution. The structure of the worksheet can be seen in Fig. 3.
Description of a Method for the Handling of Customer Needs … 345

Table 2 ASI 4-phase model

Name of the
Phase Input = What Output = How
phase
Phase 1 Product-planning Purchaser expectation Product parameters
(performance list)
Phase 2 Component- Product parameters Component parameters
planning
(given product)
Phase 3 Process-planning Component parameters Production prescriptions
(process data)
Phase 4 Production- Process data Production instructions
planning
(task and checking prescriptions)

Correlation
among the “How”-s

What? Why?
the requirements are fulfilled Do we have to upgrade?

INPUT
What?
What to Why
Why to What
the custom- Comparison
How we contribute to the with the competition
ers want it?
fulfillment of
customer needs?

OUTPUT
How much? How much?
we want to we want to achieve
upgrade? from the perspective of What? Important and critical
“How”-s for the next phase

Fig. 3 Basic structure of a QFD work sheet

4 Example for an Automotive Industry Application

The methodology of the QFD will be shown through an example. The task is to
develop a new transport vehicle under the codename “Citysprinter” for an urban
distribution center, in order to supply various shops in the city. In the followings,
346 B. Illés et al.

Table 3 Customer expectations for the Citysprinter

Characteristics Expression
Good safety • Safety of the driver
• Safety of the cargo (refrigeration, theft, safety against impairment)
Very good travel • Turning radius <3 m
characteristics
Good acceleration • From 0 to 100 km/h in 10 s
Low fuel cost • Less than 3.500 euro/year
Lower initial cost • Less than 25.000 euro
Simple storage • Up to 1500 kg accessory cargo
• Loading/unloading should be possible from three sides

the individual steps of this task will be presented in order, according to the pro-
cedure of the QFD method.

4.1 Registration of the Customer Expectations

The customer expectations represent the first “What to do” question. Above all, this
“What to do” describes the expectations of the customer. A lot of useful information
is derived from the CRM (Customer Relationship Management) system. In prin-
ciple, the historical data is stored in the CRM system, while the consideration of the
actual trends is also necessary.
Important questions regarding the registration of the customer expectations:
• Who are my customers?
• What is the importance of these customers for my company?
• What kind of wishes and expectations my customers have?
• How important these wishes and expectations are for my customers?
The potential customer of the “Citysprinter” would like to get the vehicle with
the characteristics shown in Table 3.

4.2 Evaluation of the Customer Expectations According

to Kano

The evaluation of the customer expectations will first be realized according to the
triple classification introduced by Kano [7]. Kano [7] distinguishes:
• The basic expectations of the customer (basic demands),
• The function related expectations of the customer (performance demands) and
• Innovative properties (inspirational demands).
Description of a Method for the Handling of Customer Needs … 347

Table 4 Characterization of the customer expectations according to Kano

Characteristics Expression Basic Functional Innovative
expectation expectation properties
Safety Safety of the driver X
Safety of the cargo (refrigeration,
theft, safety against impairment)
Travel Turning radius <3 m X
characteristics
Acceleration From 0 up to 100 km/h in 10 s X
Operational Less than 3.500 euro/year X
cost
Price Less than 25.000 euro X
Loading Up to 1500 kg accessory cargo X
Loading/unloading should be
possible from three sides

The lastly specified points are often the decisive demands from the point of the
actual purchase. In case of the Citysprinter, the evaluation of the customer demands
give the results shown in Table 4. From the table, it can be seen that most of the
customer expectations are directly named by the customers themselves. It can be
concluded that the innovative characteristics are missing. Under such conditions,
the “Citysprinter” would probably not be an outstanding hit among the customers.

4.3 Weighting of the Customer Expectations

Not all of the customers’ expectations are equally important; they have to be
weighted against each other. Any kind of known weighting procedure can
be applied. In our example the method of pairwise comparison is utilized (see
Table 5). The principle of the algorithm is the following: each time, two charac-
teristics are compared with each other. If one of them is more important than the
other, then it gets 2 points, while the other gets 0. If both of them are equally
important, then they both get 1 point. After that, the normalization is carried out,
where the highest value will be 10 (Table 5).

4.4 Derivation from the Technical Expectations (HOW)

In this step, the “WHAT” purchaser expectations are translated into the “HOW” for
the “Citysprinter” product. The “HOW” means the desired characteristics of the
product. We decided that the product should fulfil the state of the art technical
 348

Table 5 Weighting of the customers’ expectations with pairwise comparison and normalization on a 1–10 scale
Pairwise comparison Travel characteristics Acceleration Operational cost Price Loading Summary Normalization
matrix on 1–10
2—more important
1—equally important
0—not important
Travel characteristics 1 0 2 2 5 7.1
Acceleration 1 0 1 1 3 4.3
Operational cost 2 2 2 1 7 10.0
Price 0 1 0 0 1 1.4
Loading 0 1 1 2 4 5.7
B. Illés et al.
Description of a Method for the Handling of Customer Needs … 349

requirements in the case of the vehicle’s dimensions, the motorization, the hybrid
engine, the production costs and the payload.

4.5 Quantitative Expression of the Technical Requirements

Now the objective values can be determined according to our technical require-
ments, but it is more preferable to evaluate the technical significance and the
competitiveness first. This is the step where the connections between the product
characteristics (HOW) and the customer expectations (WHAT) are determined.
Here it is asked that how strongly each unique product criteria affect the fulfilment
of the customer expectations (correlation). The strength of the connection between
the HOW and the WHAT is evaluated with the help of a previously fixed scale.
In our example, the evaluation is carried out in four steps:
0—no connection,
1—narrow/weak connection,
2—medium connection,
3—high level/strong connection

4.6 Calculation of the Technical Meanings of the “HOW”-s

For the evaluation of the Technical meaning, the sum should be formed by taking
into account the importance and the suitable demand in relation to all products.
Based on the values, the priorities among the characteristics of our technical
solution for the “Citysprinter” is decided.
In our example, the following priorities are revealed:
1—the motorization has the highest priority,
2—the manufacturing costs has the second priority,
3—the hybrid engine has the third priority,
4—the vehicle dimensions have the fourth priority,
5—the useful loading has the lowest priority.

4.7 Comparison with the Concurrent Products (Comparison

of the “WHAT”)

Now the comparison between the own and the concurrent products is carried out.
By doing this, the own products and the similar products of the competitors are
350 B. Illés et al.

determined and listed. This creates a documentation regarding the market image of
the own and the concurrent products, while it also gives the prestige of the different
products in relation to each customer expectation. The data which are the basis of
the evaluation has to be specifically produced, or it could be provided by the sales
department based on the systematic study of the market.

4.8 Investigation of the Interdependencies Among

the Characteristics (Correlation of the HOW
to the HOW) and the Determination of the Direction
of Optimization

It should be emphasized, that the parameters are mutually strengthen (positive

correlation) or weaken each other (negative correlation). Based on the correlation,
critical constellations are revealed which should be resolved. This can be done for
example through the modification of the prescribed values (“how much”).
Now the House of Quality in phase 1 can be fully completed. The result is
shown below in Fig. 4.

4.9 Difficulty at the Realization and Technical Comparison

with the Competition

In the next step, the difficulty of the realization of the “HOW” is evaluated. Among
others, the evaluation contains the statistics, the tests, the planning, the experiences
regarding the improvements, the reclamations, the terms of the guaranteed per-
formances and for the waste disposal, the legal terms and the mode of realization of
all the previous. As can be seen from the example, the “House of Quality” provides
a very good support as a worksheet, and also for the documentation.
For the further development of our “Citysprinter”, at least three extra Houses of
Quality are needed. The first one is required for the deduction of the characteristics
of the constituents of the concrete product, the second one is required for the
manufacturing prescriptions and the third one is required for the deduction of the
process instructions (services, checks etc.).
Description of a Method for the Handling of Customer Needs … 351

Level of optimization
2 Management decision
QFD - project 1
2 1 GF decision
Citysprinter 2 1 1 -1 GF decision
1 1 2 1 -2 Management decision

Optimization o o
HOW requirement for the Evaluation of the
design
customer

Vehicle dimensions

Production costs

Service reclamations
Hybrid engine
Classification by sales

Motorization
Meaning

Payload
WHAT demand of the
customer 5 4 3 2 1
Travel characteristics 7,1 3 2 0 1 1 1
Acceleration 4,3 0 3 1 2 1 1
Operational cost 10,0 1 2 3 2 0 2
Price 1,4 1 2 1 3 1 3
Loading 5,7 2 1 2 2 3 1
l / 100

CW -
Wert
km

Legend
Meter

Measurable characteristics of the

kW

Kg

product Own product

Technical significance 44,3 55,7 47,1 51,4 30,0 A concurrent product
Priority of the QFD calculation 5 1 2 3 4 B concurrent product
technical difficulty 3 5 2 4 5 1 bad
organizational difficulty 2 5 5 3 5 5 very good
5 What/How connection strength
4 0 none
Technical competition 3 1 weak
2 2 mediocre
1 3 great

Fig. 4 The accomplished House of Quality after the close of phase 1

5 Important Matrices

The method uses two important matrices. The first is the component matrix and the
second is the process matrix. We have to examine these matrices.

5.1 Component Matrix

In the component matrix, the determination and the specification of the critical
product components is carried out. The selected product characteristics (How),
352 B. Illés et al.

their (vertical) and prescribed values (“How much”) are taken over to a new “House
of Quality” (component matrix).
For all products (all of the own products and all of the concurrent products) an
own matrix is formulated. The product functions such as stability, driving, etc. are
put vertically into this matrix (these can be regarded as “What”) and are supple-
mented by such aspects as transportation, selling and delivery. Horizontally, the
product components and aspects such as ordering, accounting and transportation are
inserted as “How”. In addition, the estimated costs of each component are num-
bered. The function costs of each component for the “From which to How” are
given in percentage, in order to have an overview of the cost structure and its
distribution. From the representation, it will be visible if for example the concurrent
products are able to satisfy the same functions with fewer components.

5.2 Process Matrix

The harmonization of the procedures and the processing of the critical process
elements, the optimal values of the process parameters, the costs and the reliability
are done in the third phase of the QFD for further follow-up. The process during the
process planning has to be determined in such a way, that the specifications of the
components in the manufacturing could be kept in a reproducible manner. Through
the optimization of the process (e.g. by statistical experimental methodology) often
principal improvements can be gained without significant extra investment.

6 Other Areas of Utilization of the QFD Method

in Logistics

As it can be seen, the QFD is a general methodology at the company in order to

maintain the continuous customer orientation. It is easy to apply it for a new task, at
a specific company and at a given sector. Inside logistics, its utilization is possible
for example in the areas shown in Fig. 5.
Description of a Method for the Handling of Customer Needs … 353

Company planning Logistics planning

Logistics networks Packaging

Strategic alliance Loading units
Loading equipment
Logistics QFD
services applications Planning of
in logistics logistics systems
Packaging, Transportation
Distribution centers
Storaging, Commissioning
Complex automation
Financing, Waste handling Customer requirements systems
for the products
Software and services Planning of
development logistics processes

Cargo management systems Order fulfillment

Spedition software Loading
Telematics software Unloading
Supply-Chain-operation
Distribution processes
Planning and building Controlling processes
Staff planning
of logistics centers
Logistics centers, Application and qualification
Significant developments

Fig. 5 Examples in the fields of application of QFD in logistics

7 Conclusions

In the paper, we presented the QFD method, which is a well refined tool for the
evaluation and quantification of the customer needs and expectations. The method
is becoming increasingly applied in the field of logistics, while it is also known in
the automotive industry. For this reason, we chose to describe the QFD through a
practical application that relates to both industries, which represents very well the
strong connection between these two important fields of engineering.
Besides the presented example, the QFD can be applied in various other fields of
logistics as well, areas that are also crucial in supporting the modern automotive
manufacturing processes. These areas include the planning of logistics systems, the
planning of logistics processes, the proper development of cargo management
systems and software, the efficient design and operation of complex supply chains,
the proper utilization of controlling processes and many other essential areas.
It is important to see that the QFD is just one tool among the multiple methods
which are utilized in the quality management of modern logistics systems. Other
important tools are for example the prevention methods utilized for the avoidance
and reduction of failures (methods such as FMEA, Fault Tree Analysis, Poke-yoke
and others), the methods of Benchmarking and Business Process Reengineering,
354 B. Illés et al.

the crucial field of Statistical Process Control, or the very well-known and widely
applied methodology of Kaizen with its various techniques. The latter is especially
important, as it forms one of the backbones of the Lean philosophy, which again
plays a vital role in the efficient operation of the modern automotive manufacturing
system [1, 8]. All of these examples show that the continuous development of the
quality management methods used in logistics has a wide and positive effect on
many other related areas, out of which one outstanding beneficiary is the auto-
motive industry.
Finally, it should be noted that such modern trends like the extensive use of big
data analysis and the rapid spread of the internet of things (trends that together
define the concept of the so called “industry 4.0”) also support the wider use of such
data-intensive techniques as the QFD. As the accumulated data related to customer
expectations grows with an almost exponential rate, the importance of these quality
management tools will grow accordingly in both the logistics and automotive
industries.

Acknowledgements “This project has received funding from the European Union’s Horizon
2020 research and innovation programme under grant agreement No 691942”. “This research was
(partially) carried out in the framework of the Centre of Excellence of Mechatronics and Logistics
at the University of Miskolc”.

References

1. Tamás P (2016) Application of value stream mapping at flexible manufacturing systems. Key
Eng Mater 686:168–173
2. Kovács Gy (2012) Productivity improvement by lean manufacturing philosophy. Adv Logistic
Syst: Theor Pract 6(1):9–16
3. Kovács Gy, Illés B (2011) Productivity improvement by application of lean manufacturing,
conference proceeding, international scientific conference (MASXXI 2011), ISBN:978-959-
250-693-0, pp 1–6
4. QFD—quality function deployment/ausgearb. von der Arbeitsgruppe 132 “Quality Function
Deployment”. Hrsg.: Deutsche Gesellschaft für Qualität e.V. (DGQ).—Berlin; Wien; Zürich:
Beuth, 2001. DGQ-Band; 13–21 ISBN 3-410-32899-8
5. Akao Y (1992) QFD—Quality Function Deployment: Wie die Japaner Kundenwünsche in
Qualitätsprodukte umsetzen. Verlag Moderne Industrie, Landsberg. ISBN 3-478-91020-6
6. King B (1994) Doppelt so schnell wie die Konkurrenz; dt. Übersetzung: Kossmann; Hofstetter;
Lange; Grohn; St. Gallen; gfmt
7. Kano N, Seraku N, Takuhashi F, Tsuji S (1984) Attractive quality and must-be-quality.
Hinshitsu: J Japan Soc Qual Control 39–48
8. Tamás P (2016) Application of simulation modeling for formation of pull-principled
production control system. J Prod Eng 19(1):99–102
Sensorless Determination of Load Current
of an Automotive Generator Applying
Neuro-Fuzzy Methods

Csaba Blága

Abstract This paper presents a sensorless method for determination of the load current
of an automotive generator applying a neural-fuzzy implementation. We developed a
simulation model of the automotive generator and its voltage regulator in order to get
information about its behaviour at different operating conditions. The model takes into
consideration the nonlinearity caused by the saturation of the magnetic flux and the
effect of the shaft speed on to the internal impedance. The simulated results are com-
pared to those that are available in the literature and to the results gained from real
system measurements. A laboratory test rig was developed to study the operation of the
automotive generator and voltage regulator in different conditions. In contrast to general
spread methods in this area the measurement results are plotted in 3D to emphasize the
hidden operation fields. In the operation of the system the parameters of the DFM—
Digital Field Monitoring—signal have important roles. Both the frequency and even
more the duty of the DFM signal carry on important information about the condition of
operation of the automotive generator and its voltage regulator, especially about the load
current. This has an important influence on the whole system of the electric circuit of a
car starting from battery to end consumers, different ECUs—Electronic Control Units—
fuel consumption and emission of the ICE—Internal Combustion Engine. Applying
neural-fuzzy theory, we could realize a sensorless method for this aim.

Keywords Automotive generator Voltage regulator
Sensorless current deter-


mination Neural-fuzzy implementation

1 Introduction

This paper presents a simulation model of a voltage regulator of an automotive

generator (or alternator) in order to get information about its behavior at different
operating conditions. The role of the voltage regulator is to keep the voltage of the

C. Blága (&)
University of Miskolc, Miskolc, Hungary
e-mail: elkblaga@uni-miskolc.hu

© Springer International Publishing AG 2017 355

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_32
356 C. Blága

alternator of the car at a constant value that it is necessary to charge the battery and
to supply the consumers. Nowadays the most spread passenger cars have an electric
circuit of “12 VDC”. It is known that even the starter (lead-acid) battery has a
greater voltage than 12 VDC. If we take into consideration that the battery has to be
charged, then we can assume that it is necessary to produce at least 14 VDC quite at
low operation speed of the internal combustion engine (for example at idling) to
avoid the discharge of the battery especially in urban traffic. The speed of an
Otto-engine can increase from about 800 rpm to 6500 rpm. If the induced voltage is
somehow proportional to the rotational speed of the shaft, then it means the voltage
would increase over 100 V and all the consumers would blow out. The voltage
regulator is set in such a way that the voltage is 14.5 VDC. So the voltage regulator
has an important role, especially nowadays, because there are many electronic
equipment and apparatus in a car. Also there are new strategies concerning to the
charging methods and operation of batteries that requires important interventions
from the voltage regulator. The voltage regulator is also controlled by the central
electronic control unit of the car through different communication lines as LIN,
CAN or FlexRay. Integration of voltage regulator into the diagnostic system of
vehicle has become a common service, also.

Fig. 1 Construction of an alternator of a car: 1 housing, 2 stator, 3 rotor, 4 transistor voltage

regulator with brush holders, 5 collector rings, 6 rectifier, 7 fan [2]
Sensorless Determination of Load Current … 357

Nowadays, special generators have already appeared on the market and built into
cars, the so called starter-generators, as it is described in paper [1]. These are
suitable to start the internal combustion engine as a starter motor and afterwards to
operate as a generator.
In order to be able to simulate the behaviour of the voltage regulator we should
know the construction and the operation of the entire system. The alternator basi-
cally is a synchronous generator as it is presented in Fig. 1 and described in many
references e.g. [2]. It can be observed that the housing contains a rectifier and a
voltage regulator, nearby other elements as brushes, fan, etc.

2 Simulation Model of an Alternator

To create the simulation model of the alternator we started from a simplified ver-
sion. An improvement of this model has been carried out by introducing into the
model the followings:
• the saturation of magnetic flux and
• the effect of rotational speed of the shaft on the internal impedance of the
synchronous generator.
The induced voltage depends on the magnetic flux. The saturation of magnetic
flux has a great influence on the operation of an alternator [3, 4]:
  
Ff Iex
U ¼ Usat sin arctg ; ð1Þ
Usat

where Usat = saturated value of flux, Ff = flux factor, Ff = U/Iex, and Iex = the
excitation current.
The induced voltage can be calculated as follows:

Vi ¼ kU x; ð2Þ

where k = machine constant, x = angular velocity [rad/s].

Moreover the frequency depends on the rpm of the shaft, so the internal
impedance also depends on the rotational speed of the shaft.
pn  p pniL
a
X ¼ x L ¼ 2p fL ¼ 2p L¼ ; ð3Þ
60 30

where L = inductance [H], f = frequency [1/s], p = number of pole pairs,

na = rotational speed of the shaft of the alternator [RPM], n = rotational speed of
358 C. Blága

Fig. 2 Simulation model of an automotive generator. Inputs: 1 rpm, 2 remanent flux, 3 internal
resistance, 4 load current, 5 excitation current, 6 time, 7 inductance of stator. Outputs: 1 line
voltage R-S, 2 line voltage S-T, 3 line voltage T-R, 4 frequency

the internal combustion engine [RPM], i = rate of the mechanical transmission

between the internal combustion engine and the alternator.
We had to build in a feedback from the output current. The output current is the
load of the alternator. There will be a voltage drop inside of the alternator and the
output voltage will decrease if the load current increases (Fig. 2).
Concerning to the external behaviour of the alternator we can make two
affirmations:
• the higher the speed is, the higher the output voltage is (taking into consider-
ation the nonlinearity caused by the saturation of the magnetic flux), and
• the higher the load current is, the lower the output voltage is.
The rectifier model is presented in Fig. 3.
A simulation model of the battery is presented in Fig. 4.
Sensorless Determination of Load Current … 359

Fig. 3 Model of the rectifier. Inputs: 1 phase 1, 2 phase 2, 3 phase 3. Output: DC voltage

Fig. 4 Model of the battery. Inputs: 1—charging voltage, 2—internal resistance of battery, 3—
time, 4—capacity of the battery, 6—internal voltage of battery. Output: 1—network voltage
360 C. Blága

3 Performance Curve

In order to have the performance curve, the alternator has to operate in such a
condition that the regulator should not have any influence on its operation. So this
would give an unregulated characteristic: the load current is changed at different
rpm in order to have constant voltage. In real cases the alternator has its regulator
built in. Maybe it is difficult, but it would be necessary to separate its influence.
Similar happens on a test bench where in the first step the level of regulated voltage
should be found. The test voltage should be set below this value, that means the
voltage regulator will not control the excitation field. In this case we have an easier
situation because the voltage regulator can be eliminated from the system. Also the
battery is removed because its internal voltage and charging current would modify
the behaviour of the alternator.
It can be observed (Fig. 5) that the output voltage (2) of the alternator was kept
constant, but the load current (1) increases and there is no any change in the
excitation current (3). The performance curve is presented in the Fig. 6.

Fig. 5 Time domain analysis of the performance curve: horizontal axes represents the time [s]; on
the vertical axes there are presented the following quantities: 1 the load current [A], 2 output
voltage [V], 3 excitation current [A]. (The scale should read as it is.)

Fig. 6 The performance

curve of an alternator by
simulation: I(n), V = const.
On horizontal axes it
presented the rotational speed
[RPM] of the shaft of the
alternator and on the vertical I [A]
axes it is presented the load
current [A]

n [RPM]
Sensorless Determination of Load Current … 361

Fig. 7 The performance

curve I(n) that can be found in
literature [2]

The simulated characteristic and the curves known from the Ref. [2] are identical
if same circumstances are assumed (Fig. 7).

4 Simulation Model of a Voltage Regulator

The voltage regulator has passed a nice trajectory of evolution since the hero era of
car driving. It started from the electro-mechanical regulator, over through-hole
technology and surface-mounted technology until to power modules and specialised
integrated circuits (Fig. 8).
At the base of each regulator stays a switching element that is a contactor or a
transistor. It works mainly as a simple delta modulator. Each time the voltage
reaches the upper and lower limits of the tolerance band around the reference value
of the voltage (14.5 V) the switch changes its state.
It results that the excitation current rises and falls alternatively. The ratio of the
switch-on and switch-off times will differ from one operating point to other. The
steepness of the slope depends on the level of the current. The switching frequency
is changing also. So the pulse width and the pulse frequency change in the same
time (Fig. 9).
On Fig. 10 we can observe the electric circuit of alternator, voltage regulator and
battery. Inside the voltage regulator there are two main components: a switching
element (1) and a comparator (2). If we would like to build the model of a voltage
regulator we have to use such kind of components. The simulation model of a
voltage regulator is presented in Fig. 11.
362 C. Blága

Fig. 8 Evolution of technology of voltage regulators: a electro-mechanical, b through-hole

mounted, c surface mounted, d power modules (smart power)

Fig. 9 Operation of voltage regulator [2]

Sensorless Determination of Load Current … 363

Fig. 10 Electric circuit of an alternator, voltage regulator and battery

Fig. 11 Model of the voltage regulator. Inputs: 1 leakage current, 2 voltage base signal, 3 network
voltage, 4 resistance of excitation, 5 frequency, 6 inductance of excitation. Outputs: 1 excitation
current, 2 DFM = Digital Field Monitoring
364 C. Blága

5 Simulation Model of Entire System

Based on the previously described elements of the system a nonlinear model of the
alternator including the rectifier and the voltage regulator can be created. Setting the
parameters of the inputs the simulation can be carried out. The rpm of the internal
combustion engine is set to change linearly from 1000 to 6000 rpm in every 50 s,
and the load resistance changes in steps from 5.5 to 0.5 X in every 10 s (Fig. 12).
The results of the simulation are presented in Fig. 13. It can be seen that the
induced voltage (1) has ripples due to the rectification and due to the voltage
regulator. The network voltage (2) is highly influenced by the charging level of the
battery and the state of the load current (3). The DFM (Digital Field Monitoring)
signal gives information about the operation mode of the voltage regulator and so
about the excitation current. A value of 5 V of DFM signal means that the switch
state is ON: so the transistor conducts. If it is zero that means the switch is OFF: so
the transistor does not conduct. The higher the duty is, the higher the excitation is.

Fig. 12 The entire model of system

Sensorless Determination of Load Current … 365

1 3
2
4

t [ms]

Fig. 13 The result of the simulation: horizontal axes it presents the time [s]; on the vertical axes
are presented the following quantities: 1—the induced voltage in the stator winding [V], 2—
voltage of the network [V], 3—load current [A], 4—DFM signal [V]

Fig. 14 Laboratory
measurement of the alternator

6 Laboratory Measurements

A special test bench (Fig. 14) was built to validate the simulated model with real
alternator. Because there was not possible to use an internal combustion engine, we
applied a DC motor that was able to drive the alternator with suitable velocity. The
speed could be changed by changing the voltage of the DC power supply. The load
current of the alternator could be set by an electronic (artificial) load.
A series of measurements were carried out using equipment presented in
Table 1.
In the first step we set the value of the load current, then changing the supply
voltage of DC motor we set certain values of the speed and finally we measured the
output voltage of the alternator. In this way we obtained the following characteristic
surfaces as you can see in Fig. 15.
366 C. Blága

Table 1 Measurement equipment

Item Model Specifications
Electronic load KIKUSUI 1000 W, 15 V
PLZ1004 W 0–200 A
Current clamp Fluke i30 DC to 20 kHz, 5 mA–30 A
conversion 100 mV/A
Digital multimeter Volcraft
Battery Perion 44 Ah, 12 V
DC motor CEM 2000 W, 169 V, 3.6 A
3000 rpm
DC power supply Built in laboratory power 0–500 V, 0–10 A
Digital oscilloscope SDS 200 A 200 MHz, 5 GS/s, max. 42 Vpk
PC Portocom Freestar

(a) the 1st alternator

17

16.5
16 16.5-17

15.5 16-16.5
15.5-16
U [V] 15
15-15.5
14.5
14.5-15
14 14-14.5
13.5 13.5-14
13 6906.06 13-13.5
4604.04
0

2302.02
8

12

I [A] n [1/min]
16

0
20

(b) the 2nd alternator

17
16.5

16 16.5-17

15.5 16-16.5
15.5-16
U [V] 15
15-15.5
14.5
14.5-15
14 14-14.5
13.5 13.5-14
13 6121.08 13-13.5
4080.72
0

2040.36
8

12

I [A] n [1/min]
16

0
20

Fig. 15 Characteristic surface of the voltage regulator in function of the load current and the shaft
speed a bed regulator, b good regulator
Sensorless Determination of Load Current … 367

In the case of the first alternator we could observe that at high speed and at high
load the voltage regulator had an interesting behaviour: the voltage increased up to
value approx. 16.5 V. This would result in a malfunction of several consumers or
even a broken off some consumers of the electric circuit of the car. In the case of the
second generator the voltage regulator worked properly.
Moreover, we could find some interesting correlations between the duty and the
frequency of the exciting current and the load current of the generator, as well as the
rotation speed of the shaft of the generator (RPM) presented in Figs. 16 and 17.

Fig. 16 Measurement results of exciting current versus time: correlations between a duty and
current, b frequency and RPM

Fig. 17 Simulation results of exciting current versus time: correlations between a duty and
current, b frequency and RPM
368 C. Blága

Fig. 18 Influence of current

100
and RPM on duty got from
simulations: duty = 50
f(I, RPM), where duty duty [%] 0

65
[%] = 100
TON
f

500

45
2500

25
4500

6500
RPM [1/min] current [A]

Fig. 19 Influence of current frequency

and RPM on frequency [Hz] [Hz]
got from simulations: 3000
f = g(I, RPM) 2500
2000
1500
6500 1000
RPM 500
3500
[1/min] 0
500 15 5
35 25
55 45
current [A]

Fig. 20 Unequivocal
dependency of current from 60
duty and frequency:
40
I = h(duty, RPM) current [A]
20
4500 RPM
0 [1/min]
100
80
60

500
40
20

duty [%]

For us Fig. 19 is less important, but changing the variables of surface presented
in Fig. 18 we get the following diagram presented in Fig. 20.

7 Adaptive Neuro-Fuzzy Implementation

The previous characteristic surface can be implemented in a neuro-fuzzy applica-

tion as follows in Figs. 21, 22, 23, 24 and 25.
Sensorless Determination of Load Current … 369

Fig. 21 Surface of current versus duty and frequency—I = h(duty, RPM)—implemented in a

neuro-fuzzy application

Fig. 22 Neural network of dependency of current versus duty and frequency I = h(duty, RPM)
370 C. Blága

Fig. 23 Block diagram for determination of current versus duty and frequency—I = h(duty,
RPM)—based on a neuro-fuzzy implementation

Fig. 24 Simulation results for determination of current versus duty and frequency—I = h(duty,
RPM)—based on a neuro-fuzzy implementation
Sensorless Determination of Load Current … 371

Fig. 25 Possibility of elimination of the electric load detector (ELD) [5] as a practical application
of determination of current versus duty and frequency—I = h(duty, RPM)—based on a
neuro-fuzzy implementation

8 Sensorless Determination of RPM

If we measure the voltage of electrical circuit of the car we can detect pulses coming
from the rectifier of the generator. The frequency of the pulses depends on the RPM
of the generator that is driven by the internal combustion engine. So, both the RPM
of the generator and internal combustion engine can be determinate from the fre-
quency of the pulses. Only we to build an electronic circuit to detect the commu-
tation point between pulses, for example. This method and circuit was developed
and built as it is presented in the literature [6]. The measurement results are pre-
sented in Fig. 26.
Determination of the RPM can be done with the following relation:
f
nALT 60fel 60
Pp 5
fp ½Hz
nICE ½min1  ¼ ¼ ¼ ¼ ; ð5Þ
i p
i p
i 3
i
372 C. Blága

Fig. 26 Detection of commutation point of pulses of voltage of electric circuit of car

where
nICE = RPM of internal combustion engine, nALT = RPM of alternator,
i = ratio of the belt transmission, fel = frequency,
p = pole pairs of alternator (= 6), fp = frequency of pulses,
P = number of pulses (6-pulse Graetz-bridge).

9 Conclusions

This paper contains a lot of theoretical and special knowledge on the field of
electrical machines and power electronics, mathematical deductions, simulation of
linear and non-linear systems, building of test equipment and interpretations of
many measurement results. The main aims have been achieved: the simulation
model has been built and measurements were done. The simulation model of the
voltage regulator has been created. It can be stated that the simulated behavior of
the voltage regulator corresponds to the expectations and to the results of the
laboratory measurements. We can simulate the operating conditions of the power
electronic semiconductor device that switches the excitation current of a car
alternator. As a continuation of this work we can imagine to build simulation
models based on soft-computing methods as fuzzy-logic, neural network or quite
use an adaptive neuro-fuzzy inherence system.
Sensorless Determination of Load Current … 373

Acknowledgements This research was carried out in the framework of the Center of Excellence
of Mechatronics and Logistics at the University of Miskolc.

References

1. Viorel A, Szabó L, Löwenstein L, Şteţ C (2004) Integrated starter-generators for automotive

applications. Acta Electrotehnica 44(3):255–260
2. Bauer H (2003) Alternators and starter motors. Robert Bosch GmbH, Stuttgart
3. Blága Cs, Kovács E (2003) Solenoid based actuators, EDPE’03. In: Proceedings of the
international conference on electrical drives and power electronics. The high tatras, Slovakia,
24–26 Sept 2003, pp 207–210
4. Blága Cs, Kovács E (2011) Modeling and measurement of an alternator. In: Proceedings of the
17th international conference on electrical drives and power electronics. The high tatras, Stara
Lesna, Slovakia, 28–30 Sept 2011, pp 210–214. ISBN 978-80-553-0734-3
5. Veres L, Analysis of intelligent generators, Autotechnika, X-Meditor Kft., Győr, 2014/4. (in
Hungarian)
6. Domonkos D (2014) Measuring of the internal combustion engine speed using electrical
interference, final thesis, Miskolc (Hungary), 2014. (in Hungarian)
Distributed CAN-Bus Based Driving
Assistance System on Autonomous Vehicle

Gergely Kovács and László Czap

Abstract This article focuses on CompactRIO based driver assistance system and
the goal was a completely autonomous operation. By the development our vehicle
had to be able to realize several intelligent driving assistance functions, such as
adaptive cruise control, lane keeping, predictive emergency braking, brake energy
regeneration, automated parallel- and cross parking and GPS navigation, and we
also had to design hybrid drive on the go-kart. For implementing the intelligent
functions mentioned above, we chose NI cRIO during the development. By taking
the advantages provided by the developing environment and the modularity of the
system, we could solve the scheduled tasks.

1 Introduction

The Hungarian Bosch Group organized a series of competitions in 2013 for the
engineering universities of Hungary. The goal was to design and create a driving
assistant system for a small racing car, for a go-kart which was provided by the
company itself. The competitors were also forced to use several and existed Bosch
components, such as an MPC (Multi Purpose Camera) camera, a ParkPilot ECU
(Electronic Control Unit) with its ultrasonic sensors and a Bosch Mid-Range radar
(MRR). During the preparation phase the individual teams had to fulfil special
milestones in order to achieve minor goals and allowed the Bosch Company to
monitor and support the teams. The final aim of the competition is to develop a
smart driving system not just for go-karts but for passenger car, to create a basis for
an autonomously driving vehicle.

G. Kovács (&)
L. Czap

University of Miskolc, Miskolc, Hungary
e-mail: kovacs.gergely17@gmail.com
L. Czap
e-mail: czap@mazsola.iit.uni-miskolc.hu

© Springer International Publishing AG 2017 375

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_33
376 G. Kovács and L. Czap

2 The RIO Architecture

The mechanical impacts affected on the go-kart, such as vibrations or knocks, the
thermal stress and the problems come up, especially from outside, the environ-
mental damages require an industrial designed platform. CompactRIO’s modular
I/O enables users to flexibly implement a wide array of sensor types and industrial
connectivity. The CompactRIO real-time controller and reconfigurable FPGA
chassis provide a customizable inherent ability to implement many parallel loops on
the FPGA (Field Programmable Gate Array) and real-time controller, its rich set of
complex math functions makes it powerful software to control all aspects of a smart
machine. Additionally, the open and flexible nature of LabVIEW helps implement
reliable communication architectures for local operator interfaces and centralized
resource management, diagnostics, and remote updates. These functions are
essential once the machines are deployed around the world and need maintenance.

2.1 Real Time Controller

The LabVIEW Real-Time Module (Fig. 1) combines LabVIEW graphical pro-

gramming with the power of a real-time operating system, enabling you to build
deterministic real-time applications. National Instruments designed the LabVIEW
Real-Time Module to execute VIs on three different real-time operating systems.
The LabVIEW Real-Time Module can execute VIs on hardware targets running the
real-time operating system of the NI Embedded Tool Suite (ETS), Wind River
VxWorks or NI Linux Real-Time. The dataflow programming model of LabVIEW
frees you from the sequential architecture of text-based programming languages.
Additionally, LabVIEW makes it easy to assign thread priorities with the Timed
Loop structure. Each loop can contain a separate task with a configurable timing
source, period, priority, and more. LabVIEW Real-Time supports multicore pro-
cessing and automatically maps parallel sections of your code to individual OS
threads, which eliminates the need to manually create and manage them. By default,
these threads are also automatically balanced across the CPUs available on your
real-time hardware [1].
We have chosen the Linux Real-Time operating system, because it provides a
wide range of opportunities to connect other devices to our system. NI Linux
Real-Time contains embedded user interface and Mini Display Port to implement a
local HMI (Human Machine Interface).
Distributed CAN-Bus Based Driving Assistance System … 377

Fig. 1 Real-time development tools

2.2 FPGA Module

The basic components of a LabVIEW FPGA system are:

• the LabVIEW FPGA Module,
• the NI-RIO driver,
• and an NI-RIO device, also known as RIO target.
The LabVIEW FPGA Module allows LabVIEW to target FPGAs (Fig. 2) on NI
Reconfigurable I/O (RIO) hardware, so scientists and engineers can take advantage
of the performance and flexibility of FPGAs without the needing to learn low-level
design tools. The LabVIEW FPGA Module compiles the LabVIEW VI to FPGA
hardware, the graphical code is translated to text based VHDL code. LabVIEW is

Fig. 2 NI FPGA architecture

378 G. Kovács and L. Czap

well suited for FPGA programming because it clearly represents parallelism and
dataflow, so users who are both experienced and inexperienced in traditional FPGA
design can productively apply the power of reconfigurable hardware. The Xilinx
compiler tools synthesize the VHDL code into a hardware circuit realization of the
LabVIEW design. The result is a bit-file that is loaded to the FPGA chip before
running the application [2].

3 The Control System

By choosing our control unit we had to take the external environmental impacts and
the software development into consideration. During the two years we had to
implement numerous driver assistances, so we needed a control unit which not only
has the suitable industrial environment, and the algorithms can be implemented on a
high level. Because of this development time can be drastically reduced, compared
to lower level programming languages. During the operation of the go-kart,
mechanical effects like vibrations, heat effects and other environmental effects occur
more often, so industrial protection is a requirement. After taking these into con-
sideration, we used the CompactRIO platform (Fig. 3) from National Instruments
from the beginning, which are fully suitable for us. Its compact and provides robust
protection against the environmental effects, the operating system running on the
controller provides the time critical development.
In the first year we used the 9022 controller with the 9012 FPGA provided by
National Instruments. Because of the complexity of the task and the enormous
number of control tasks, it was necessary to use a more powerful CompactRIO
platform, because we needed a larger FPGA and better computing capacity of the
control system. Our other problem was the lack of space of electric components and
because of the lack of ventilation we needed a platform, whose heat production is
lower. This is the reason why we chose National Instruments 9031 CompactRIO,
where the FPGA was given. It also played a role, that the controller is using the

Fig. 3 NI 9031 CompactRIO

Distributed CAN-Bus Based Driving Assistance System … 379

Real-Time Linux operating system, so other devices can be easily connected to the
CompactRIO.

3.1 I/O and Communication Modules

We had to choose the modules we would like to use during the development. It was
important that we only had 4 slots available, so we preferred multifunction mod-
ules. Two multifunction modules were selected: a digital I/O module with 32
channels and a universal analog input module. We chose the following modules:
NI 9403 Bidirectional Digital I/O
• 32 digital I/O
• 5 V TTL level
NI 9219 Universal Analog Input
• 4 channels universal module
• 100 S/s per channel simultaneous inputs
NI 9263 Analog Output
• ±10 V output range,
• 16-bit resolution
NI 9853 High-Speed CAN Module for NI CompactRIO
• 2 ports

4 Distributed CAN-Bus System

CAN communication implements with the NI 9853 High Speed module. The main
reasons why we chose this module are,
• High Speed CAN.
• Two ports (CAN0, CAN1), two separated CAN networks with different speeds.
• Support of 11 and 29 bit arbitration field.
The only difference between the two ports is that only one of them needs
external power supply, which we can provide with 5–30 V DC power. Each port of
the NI 9853 has an NXP SJA1000 CAN controller that is CAN 2.0B-compatible
and fully supports both 11-bit and 29-bit identifiers. Each port also has an NXP
TJA1041 High-Speed CAN transceiver that is fully compatible with the ISO 11898
standard and supports baud rates up to 1 Mbps.
380 G. Kovács and L. Czap

The most important factor of our choice was that the communication functions
can be easily accessible from the FPGA [3].

4.1 Developing of the CAN-Bus System

The first step was to decide what and how fast the devices are going to commu-
nicate with CAN.
This was a problem, because other devices which are not using this timing won’t
be able to properly communicate on that network. For this reason the Park
Pilot ECU was moved to CAN1. The camera and the radar are communicating with
a transfer rate of 500 kb/s (Fig. 4). These rates are not changeable, so the AHRS
system and the microcontroller in the steering wheel are going to communicate on
that rate of the CAN0 port. We can set the speed of our network, and we can create
a so called “white list”, where we can define what kind of messages we would like
to receive [4, 5].

4.2 Communication with the Park Pilot ECU

The Park Pilot ECU was the first Bosch component we had to fit on the CAN
network. Bosch gave the specifications about the network and the messages. The
Park Pilot ECU is the only device on a separate network as mentioned above. There
are twelve ultrasonic sensors connected to the Park Pilot ECU, which signals are

Fig. 4 CAN-BUS system

Distributed CAN-Bus Based Driving Assistance System … 381

fitted with temperature and humidity compensation. The sensor data arrive in four
messages, so three sensors send their data in one message:
• eight of the sensors can see up to 2.5 m,
• four sensors can see up to 5 m.
We decided that we position the four longer range sensors on the corners of the
go-kart which was important because of the parallel parking and the home envi-
ronment recognition [4, 6].

4.3 Communication with the Radar and the Camera

In the middle of the second year Bosch gave two other devices: a mid-range radar
and an intelligent lane and object information providing camera which transmits its
data through the CAN bus. Bosch gave a high speed CAN gateway for the devices,
which directly connects to our system. For proper functioning it needs the steering
angle, the speed in [m/s] and the velocity of the go-kart along the vertical axis (Yaw
rate). Both devices are communicating through the CAN0 port [7–9].

5 Summary

We closed all three years very successfully. Every year our team was the absolute
winner of the completion and in the first year we won the first prize in the best
technical solution (Fig. 5).

Fig. 5 Awards
382 G. Kovács and L. Czap

In the second year we were the absolute winner of the competition and we won
the “Best Teamwork” and the “Smartest Go-Kart” awards too. In the third year
beside the absolute first prize, we have got the first prize in the self-driving
category.

Acknowledgements This project has received funding from the European Union’s Horizon 2020
research and innovation programme under grant agreement No 691942. This research was partially
carried out in the framework of the Center of Excellence of Mechatronics and Logistics at the
University of Miskolc.

References

1. National Instruments: LabVIEW Real-Time 1 Course Manual, Sept 2012

2. National Instruments: LabVIEW FPGA Course Manual, Aug 2012
3. Trohák A, Végh J, Kovács G, Forgács Zs (2015) Implementations of driving assistance
systems using labview based platform. Proceedings 16th international carpathian control
conference (ICCC), pp 548–551
4. Trohák A, Illés B, Bíró Z (2013) Can message filter algorithms for remote diagnostics of
vehicles. Appl Mech Mater 309:213–220
5. Ajtonyi I (2008) Industrial communication systems I. AUT-INFO Kft., Miskolc, (Hungary),
p 431. ISBN: 978-963-06-5813-3, HU ISSN: 1789-5456 (in Hungarian)
6. Trohák A, Kolozsi-Tóth M, Rádi P, Méhes L, Biró Z (2011) The development of a remote
diagnostic system for vehicles. Adv Logistic Syst 5:216–220. HU ISSN 1789-2198
7. Schoettle B, Sivak M (2015) Road safety with self-driving vehicles: general limitations and
road sharing with conventional vehicles, Report No. UMTRI-2015-2. Michigan, USA
8. Schoettle B, Sivak M (2015) Potential impact of self-driving vehicles on household vehicle
demand and usage, Report No. UMTRI-2015-3. Michigan, USA
9. Litman T (2015) Autonomous vehicle implementation predictions: implications for transport
planning, transportation research board 94th annual meeting, Washington DC, USA
Wall and Object Detection with FRI
and Bayes-Classifier for Autonomous
Robot

Roland Bartók, Ahmed Bouzid, József Vásárhelyi and Márton L. Kiss

Abstract To detecting the nearly objects and walls is very important for a mobile
robot or an autonomous vehicle to collision avoidance. It needs many of sensors for
wide range detection. The results are better with more type of sensors for example
infrared sensors, ultrasonic sensors, laser distance sensors. Because of the number
of sensors it needed to use sensor fusion. It is a relatively easy way by using the FRI
based Behaviour Description Language or the Bayes-classifier.

1 Introduction

There are many expressions in human language which are difficult to translate
(compile) in computer code. These expressions make the human communication
easy. To describe the distance between two objects one can use expressions like:
“The two objects are far from each other”; “The two objects are relatively close to
each other”. Or another example is how one expresses the speed of a moving object:
one can use the words: “fast”, “slow” and “much slower”.
To translate these features, which can describe a moving autonomous robot
fuzzy logic and fuzzy sets are used. In fuzzy logic to describe these features is used
the interval [0, 1], instead of using the logic values 0 (false) and 1 (true). Fuzzy sets
use this interval to characterise the membership functions. The members of sets
have a membership function which shows what extent is part of a sets. Fuzzy sets
were first described by L. A. Zadeh in its publication “Fuzzy sets” (see [1, 2]).

R. Bartók (&)
A. Bouzid
J. Vásárhelyi
M. L. Kiss

University of Miskolc, Miskolc, Hungary
e-mail: qgeroli5@uni-miskolc.hu
A. Bouzid
e-mail: qgebouzid@uni-miskolc.hu
J. Vásárhelyi
e-mail: vajo@uni-miskolc.hu
M. L. Kiss
e-mail: autlkiss@uni-miskolc.hu

© Springer International Publishing AG 2017 383

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_34
384 R. Bartók et al.

1.1 Fuzzy Rule Interpolation

In fuzzy the knowledge is represented by the rulebase. Rulebases are a set of rule
collection, i.e. they are “If … then” type sentences, which are sets of observations
and conclusions.
In classical fuzzy logic one needs a rule for every observation. This can result in
the description of a control system with a large number of rules which implies a
rulebase with great dimension. A complete description of a system, which contains
all the rules is called full rulebase.
There are some observations which cannot described by a rule (have no rule). In
this case there is no conclusion for that observation. The missing conclusions can be
computed from their neighbourhood using existing rules. This method is called
fuzzy interpolation, based on the description of the system behaviour. In [3] was
described the Behaviour Description Language, which uses the FIVE (Fuzzy
Interpolation in Vague Environment) method described in [4]. The FIVE method
makes unnecessary the classical steps of a fuzzy control system design, such as
fuzzification and defuzzification step. In this method defuzzification became
unnecessary.

1.2 Behaviour Description Language

Complicated behaviour models are created in a simple way with Fuzzy Rule
Interpolation (FRI) based fuzzy automaton. The fuzzy automaton has got its own
declarative description language. This language is based on ethology behaviour
description languages, which already exist, most of them are built from primary
declared samples by observing the reactions and behaviours of living creatures.
The simplicity was important in the design of the behaviour description language
[3]. The language is easy to use with minimal programming knowledge.
The behaviour engine is a fuzzy automaton where the state is a vector of
membership values. The state changes are controlled by fuzzy rulebase. The
observations and conclusions are continuous values. The knowledge representation
is based on rules for modelling behaviours. A sequence of events can be described
with continuous values and continuous states. This automaton is computed in
discrete time, but its model is continuous in states.
The Description Language contains keywords and user-defined tags, which
helps to design an easy to read and easy to modify behaviour model of a given
system. The list of keywords is as follows [3]:
• rulebase: The main base elements of behaviour descriptions are the rulebases. It
contains the antecedents (observes) and consequents.
• universe: Defines the symbols and value range. Contains the user-defined
symbols and value pairs. The universe can define the antecedents and conse-
quents too.
Wall and Object Detection with FRI and Bayes-Classifier … 385

• description: Optional keyword for comments and documenting.

• rule, when, is, and: Keywords for compose a rule. The syntax is the follow: rule
“output symbol” when “antecedent symbol” is “antecedent value” end.

2 Naive Bayes-Classifier

The Bayes-classifier is a simple, scalable classifier to determine class label for

attributes. The Bayesian classifiers can be used for text categorization such as spam
filter or in medical diagnostics. The naive Bayes–classifier supposes that the
attributes are independent. The method is based on the conditional probability. The
Bayes-theorem is:

PðAjBi ÞPðBi Þ
PðBi jAÞ ¼ ; ð1Þ
PðAÞ

where i is the number of classes, Bi, A is a vector which contains the observed
attributes. In this paper is considered that the classes show “where are the walls
around the mobile robot”. The observed attributes are the measured data received
from infrared sensors. The selected classifier will be the one where the probability
P(b|A) is maximal for the A attribute vector [5].

3 The Mobile Robot and Its Environment

The mobile robot is equipped with two wheels with differential drive. The auton-
omous robot is equipped with several different sensors, such as: incremental phase
encoder for each shaft, six-axis (Gyro + Accelerometer) MEMS, infrared
diode-phototransistor pairs for distance measurement. The robot shape and sensor
layout are shown on the Fig. 1.
The robot moves in maze-like corridors and creates maps and also navigating.
The shape of maze is square, build from square shaped cells (Fig. 2). The cells
dimension is 90 * 90 mm2, wall height is 25 mm, and the size of maze is 5 * 5
cells. The robot knows the size of the maze, but not where the walls are situated.

3.1 Data Collection from the Environment

There were defined 7 cases in the robot movement. Using Bayes-classifier, a lot of
training data were collected for every case. Some movements of data collecting
method are shown on Fig. 2. The mobile robot moved backward and forward in the
386 R. Bartók et al.

Fig. 1 Mobile robot. Infrared

LEDs are represented in blue.
Measurement area of sensors
is 10°

Fig. 2 Data collection

method. For example
a corridor, b front-right,
c left-front-right

target area while collecting the measured values. The defined cases were as follows
with their probability in the train maze:
a. wall on front (P = 0.08)
b. wall on left (P = 0.12)
c. wall on right (P = 0.12)
d. wall on left and right (P = 0.24)
e. wall on front and left (P = 0.36)
f. wall on front and right (P = 0.36)
g. wall on front and left and right (P = 0.24)
Wall and Object Detection with FRI and Bayes-Classifier … 387

The probability for the above situations is calculated with the formula:

P ¼ ðnumber of possible casesÞ=ðnumber of all casesÞ ð2Þ

where the number of all cases means the robot can enter in one cell from 4
directions; the number of possible cases were defined above (a–g).
The mobile robot was moving in each case, while collecting the measured data.
The listed cases become classes for Bayes-classifier and the first 3 cases become
classes in behaviour description.

4 Behaviour Description for Wall Detection

The Behaviour description gives 3 classes: wall detected on the left, front or right.
The given value range is [0, 1] for each class. If the returned value for each above
situations is higher than 0.6 this represents an obstacle (wall) in front of the moving
robot. The binary input value range is [0.4096] according to analogue digital
converter, where the 0 means the wall is very far and the 4096 means the wall is
very close.
The universe defined for all sensors are as follows. The given example below is
for wall detected on the left back IR sensor:
universe “leftBack”
“far” 0 0
“middle” 100 0.7
“close” 720 1
“very close” 4096 1
end
The first parameter of symbol is an element of input range; the second parameter
is the membership value. This means the input characteristics runs as described, 0
when the wall distance is far and the close and very close gives 1. The output
universe is as follows:
universe “isWallOnLeft”
“no” 0 0
“yes” 1 1
end
This is the consequence of rulebase for wall detected on the left side of the robot.
It gives 0 if no wall and gives a value larger than 0 when wall detected is
approaching. If the calculated value returned by the rulebase is converted into
Boolean value when True if obstacle occurs, false if no obstacle. This value and the
values in the input universes results of the experimental measurements.
388 R. Bartók et al.

The example rulebase for wall detected on the left is the following:
rulebase “isWallOnLeft”
Rule
“No” when “leftBack” is “far” and “left Front” is “far”
End
Rule
“Yes” when “leftBack” is “middle” and “lefFfront” is “middle”
End
End
The right sensor data is not used, because generated errors in the cases d–g. The
number of rules must be equal with the number of user-defined symbols in the
universe.

5 Bayes-Classifier for Wall Detection

The number of collected data for listed classes was enough for statistical analysis.
The occurrence of each value from 0 to 4096 were counted to the defined ranges,
for example from 0 to 30, from 31 to 60, this was needed for noise reduction. Then
probability was computed for every range. This training data were used for
Bayes-classifier. A sample of distortion calculated is shown in Fig. 3.

100
80 Left back
60
40 Left front
20 Right back
0
30 60 90 120 150 180 210 240 270 300 330 360 390 420 450 Right front
0 31 61 91 121 151 181 211 241 271 301 331 361 391 421

Fig. 3 Calculated probabilities for cluster values resulted from the different sensors for “wall in
front and left and right” class (sensors represented on: green, violet, blue, orange)
Wall and Object Detection with FRI and Bayes-Classifier … 389

Fig. 4 The situation when

the Bayes-classifier gives
better result. Thin wall caused
a false detection

6 Conclusions

Both methods succeeded to detect the walls around the mobile robot. The FRI
method was not tuned. The Bayes-classifier was trained with collected data. The
Bayes-classifier could detect the walls earlier.
The difference of distance between the FRI method and Bayes-classifier was
around 5–8 mm from the centre of the cell. When the mobile robot moved too close
(5–10 mm) to wall both of methods generates false detection. The Bayes-classifier
gave better result in the situation shown in Fig. 4. Further research is needed to
implement the FRI and Bayes-classifier in FPGA using VHDL.

Acknowledgements The research work was (partially) supported by the Hungarian Scientific
Research Found grants OTKA 29326 and Fund for the Development of Higher Education FKFP
8/2000 project. This research was (partially) carried out in the framework of the Center of
Excellence of Mechatronics and Logistics at the University of Miskolc.

References

1. Kovács S (1993) Fuzzy logic control, M.Phil. theses, Technical university of Budapest, Faculty
of informatics and electrical engineering, Branch of Computer Science, Budapest (Hungary),
p 116
2. Zadeh LA (1965) Fuzzy Sets. Inf Control 8(3):338–353
3. Piller I, Vincze D, Kovács S (2015) Declarative Language for Behaviour Description.
Emergent Trends Robot Intell Syst 316:103–112
4. Kovács S (2006) “Extending the fuzzy rule interpolation “FIVE” by fuzzy observation”,
Computational intelligence, Theory and applications: 9th international conference on
Dortmund Fuzzy Days. 802 p, Dortmund (Germany), 2006.09.18-2006.09.20. Berlin;
Heidelberg: Springer-Verlag, pp 485–497. ISBN:978-3-540-34780-4
5. Downey AB (2014) Bayesian statistics made simple. Green Tea Press, p 194. ISBN:
13-978-1449370787
Optimal Formation of Logistics Networks

György Kovács, Zoltán Varga and Márk Mihalik

Abstract The growing market globalization, increasing global competition, more

and more complex products require new technologies, methods and processes. The
shorter life cycle and the complexity of final products and new customer require-
ments require efficient operation of supply chains (SC). Optimization of supply
chains results new models, concepts of value chains and new organization and
cooperation forms of members. Reduction of total cost and lead time of the chains,
and higher customer service level can be the objectives of the supply chain
optimization.

1 Supply Chain, Virtual Organization, Virtual Enterprise

Stevens [1] defined the supply chain as a system whose constituent parts include
material suppliers, production facilities, distribution services and customers linked
together via a feed forward flow of materials and feedback flow of information.
Due to fast changing market environment, globalization and global competition,
supply chains have become more and more complex networks. Reduction of total
cost and lead time of the chains, and higher customer service level becomes the
most common used objectives for operation of supply chain networks.
A virtual organization [2] is a short-term form of cooperation among legally
independent co-producers in a logistics network of long-term duration of potential
business partners for the development and manufacturing of a product. This is true
for procurement and production, as well as for product and process innovation.
Co-producers produce the service on the basis of mutual values and act towards the

G. Kovács (&)
Z. Varga
M. Mihalik

University of Miskolc, Miskolc, Hungary
e-mail: altkovac@uni-miskolc.hu
Z. Varga
e-mail: altvarga@uni-miskolc.hu
M. Mihalik
e-mail: miha02@freemail.hu

© Springer International Publishing AG 2017 391

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_35
392 G. Kovács et al.

third party as a single organization. Each co-producer is active within the area of its
core competence. The choice of a co-producer depends on the co-producer’s
innovative power and its flexibility to act as a partner in the logistics network [2].
The strength of virtual organizations is their ability to form quickly and gain
competitive advantages.
Camarinha-Matos [3] interpreted the virtual enterprise (VE) as a temporary
alliance of enterprises that come together to share their skills, core competencies,
and resources in order to better respond to business opportunities, and whose
cooperation is supported by computer networks.
Gunasekarana et al. [4] defined that virtual enterprises are characterized by
several strategic objectives:
• maximizing flexibility and adaptability to environmental changes,
• developing a pool of competencies and resources,
• reaching a critical size to be in accordance with market constraints, and
• optimizing the global supply chain.
Virtual organizations are used in more and more industries, e.g. fashion industry,
food industry, automotive industry, etc.
Figure 1 shows supply chain networks and a virtual enterprise as a temporary
alliance of enterprises. The network includes a large number of customers, pro-
duction companies and service providers. Customers can be consumers, end-users,
etc. Production companies are the final assemblers, primary-, secondary- …sup-
pliers and raw material suppliers.

Raw material Secondary Consumers/

Primary suppliers Final Assemblers
suppliers suppliers End-customers

C
S S
C

SP SP SP
RS FA SP C
S S

RS SP SP SP FA SP C
S S

RS S S
SP FA C
SP SP SP

C
S S
C

Service providers Service providers Service providers Service providers

Material flow
Virtual Enterprise (VE) Information flow

Fig. 1 Supply chain networks and a virtual enterprise

Optimal Formation of Logistics Networks 393

Possible service providers in the supply chains can be the followings:

• Logistics service providers (forwarding comp., warehousing comp., SC
managing comp., …),
• Information service providers (IT companies, telecommunication companies, …),
• Research and Development service providers (Research institutes, Universities,
Consulting enterprises, …),
• Financial service providers (Banks, Lease brokers, Consulting enterprises, …).

2 Network Optimization in Case of a Virtual Enterprise

The objective in the optimization of a supply chain network is to produce and

deliver finished products to end consumers in the most cost effective and timely
manner and to achieve higher customer satisfaction. Supply chain management and
supply chain optimization appear frequently in the literature. Large number of
publications has been also presented in the literature on topic of virtual enterprise
concepts and characteristics [5–10].
This study shows the optimization method for the operation of a virtual network.
Objective function and constraints will be elaborated.
Total cost should be minimized, taking into consideration constraints relating to
production and service capacities, inventories and flexibility in the chain.
In our theoretical problem (Fig. 1) the followings are given: products types and
quantities ordered by customers, possible suppliers of components, possible final
assemblers, possible service providers, list of required services fulfill the customer
satisfactions, manufacturing capacities of production plants, capacities at service
providers, distances between the possible objects of the network, cost and time of
activities required for manufacturing of components and final products to fulfill the
customer orders.
The optimal formation of a virtual enterprise (optimal variation of members of
the supply chain) is searched.
Indices are used in mathematical formulation are the followings: i—products; j
—possible suppliers (including raw material suppliers, primary-, secondary-, etc.
suppliers); k—final assembler companies; l—customers; m—possible service pro-
viders; t—time periods.

2.1 Objective Function—Total Cost

Total cost is including the raw material and component costs, production costs,
transportation costs, inventory costs, cost of activities of service providers and
operation cost of the virtual enterprise.
394 G. Kovács et al.

O1 ¼ Cmat þ Cprod þ Ctran þ Cinv þ Cserv þ Coper ½Euro: ð1Þ

2.1.1 Raw Material and Component Cost

The total material cost is the sum of material costs at suppliers and material cost at
final assembler:

cmat ¼ cmij
Vijt þ cmik
Vikt ½Euro; ð2Þ

where: cmij —unit material cost of raw materials and components of product i at
supplier j [Euro/piece]; cmik —material cost of product i at final assembler
k [Euro/piece]; Vijt —production volume of raw materials and components of pro-
duct i at supplier j in each t time period [piece]; Vikt —production volume of product
i at final assembler k in each t time period [piece].

2.1.2 Production Cost

The production cost (including additional activities joining to manufacturing

activity, e.g. machine setup, material supply, loading unit formation, …) is the sum
of production costs at suppliers and production cost at final assembler company:

cprod ¼ cpij
Vijt þ cpik
Vikt ; ð3Þ

where: cpij —unit production cost of raw materials and components of product i at
supplier j [Euro/piece]; cpik —unit production cost of product i at final assembler
k [Euro/piece]; Vijt —production volume of components of product i at supplier j in
each t time period [piece]; Vikt —production volume of product i at final assembly
k in each t time period [piece].

2.1.3 Transportation Cost

The transportation cost (including additional activities joining to transport activity,

e.g. loading in, loading out, loading unit formation, etc.) is the sum of transportation
of each product in each relation of the network, between suppliers and final
assembler and between final assembler and customers:

ctran ¼ ctijk
Vijkt þ ctikl
Viklt ð4Þ

where: ctijk —unit transportation cost of raw materials and components of product
i from supplier j to assembler k [Euro/piece]; ctikl —unit transportation cost of pro-
duct i from final assembler k to customer l [Euro/piece]; Vijkt —flow of components
Optimal Formation of Logistics Networks 395

of product i from supplier j to assembler k in each t time period [piece]; Viklt —flow of
components of product i from final assembler k to customer l in each t time period
[piece].

2.1.4 Inventory Cost

The inventory cost is including the holding and storage costs of stocks at suppliers,
at assembler, at customers and in some cases at service providers (warehousing
service providers):

cinv ¼ ciij
Iijt þ ciik
Iikt þ ciil
Iilt ð5Þ

where: ciij —unit inventory cost of raw materials and components of product i at
supplier j [Euro/piece]; ciik —unit inventory cost of raw materials and components
of product i at assembler k; ciil —unit inventory cost of product i at customer
l [Euro/piece]; Iijt —inventory of components of product i at supplier j in each t time
period; Iikt —inventory of components of product i at assembler k in each t time
period [piece]; Iilt —inventory of product i at customer l in each t time period
[piece].

2.1.5 Cost of Activities at Service Providers

Activities of service providers are including the processes required for financing,
manufacturing, warehousing, packaging, etc. of product i.

Cserv ¼ Cspimt ð6Þ

where: Cspimt —cost of activities of service providers required for manufacturing

and delivery of product i in each t time period.

2.1.6 Operational Cost of the Virtual Enterprise

Operational cost of the VE (Coper ) is including the management cost of the network,
cost of the information and communication technology and all of additional costs
which are required for the optimal operation of the virtual network.

Coper ¼ Coper ðncomp ; pÞ ð7Þ

This cost component is depending on the size of the network (ncomp ) and profile
of member companies (p).
396 G. Kovács et al.

2.2 Constrains

2.2.1 Production and Service Capacity Constraints

If the product i is allocated to supplier j during time period t, the production volume
(Vijt ) should be limited by minimum and maximum volume at suppliers:

Vijtmin  Vijt  Vijtmax ð8Þ

If the product i is allocated to final assembler k during time period t, the pro-
duction volume (Vikt ) should be limited by minimum and maximum volume at final
assembler:
min
Vikt  Vikt  Vikt
max
ð9Þ

If the services required for production of product i is allocated to service pro-

vider m during time period t, the activity capacity (Cimt ) should be limited by
minimum and maximum volume at service providers:
min
Cimt  Cimt  Cimt
max
ð10Þ

2.2.2 Inventory Constraint

Based on lean philosophy the inventory is one of the biggest waste in the supply
chain, but it provides the flexibility of the chain. Depending on the inventory
strategy of the supply chain, the volume of inventories at manufactures and service
providers should be limited:
min
Iijt  Iijt  Iijt ; Iikt  Iikt  Iikt
max min
; Iimt  Iimt  Iimt
max min max
ð11Þ

2.2.3 Flexibility Constraint

The virtual enterprise is characterized by a dynamic and flexible network, so the

following flexibility constraints are defined for the chain members:
• flexibility of the manufacturing system at the manufacturing companies of the
chain,
• flexibility of service providers of the chain,
• flexibility of the IT infrastructure at the companies of the chain,
• liquidity of the chain members,
• flexibility due to the organizational structure of the chain members.
These constraints are not easy to defined, but these can be given by a value in a
given interval (1–5). Number 5 represents the most flexible value.
Optimal Formation of Logistics Networks 397

3 Software Conception for Optimal Network Design

Based on the above mentioned optimization concept our research team developed a
software application for the optimization of a virtual enterprise network.
The software is written in Java programming language. Java is a general-purpose
computer programming language that is concurrent, object-oriented, and
class-based. Java is a platform free programming language. This is a big advantage
comparing other languages, it means our written software is independent to oper-
ating systems [11].
We used Eclipse software framework to develop the software. Eclipse is an open
source integrated development environment (IDE), one of the most widely used
Java IDE. It contains a base workspace and an extensible plug-in system for cus-
tomizing the environment. Eclipse is written mostly in Java and it is primary used
for developing Java applications, but it may also be used to develop applications in
other programming languages [12].

3.1 Numerical Example for Formation of Optimal

Supply Chains

In this chapter we would like to show the application and operation of a software
developed by our research team.
In our example (Fig. 2) the supply chain includes one final assembler (FA), four
primary suppliers (S11, S12, S13, S14) and five possible secondary suppliers (S21,
S22, S23, S24, S25).
Relations of possible suppliers can be seen in Table 1, the distances of different
members can be found in Table 2.
We assume that the unit production cost (cpi) of our fictive product (product A)
is 5 Eur/pieces in Europe and in America and 2.85 Eur/pieces in Asia. We also

4.
5. FA 1.
2.
5. 2. 1.
S14 3. S11
S25 S22 S12 S21
3.

4. Final assembler
S23
5. S13 Primary supplier
3.
Secondary supplier

S24

Fig. 2 Possible supply chain combinations

398 G. Kovács et al.

Table 1 Relation matrix

FA S11 S12 S13 S14 S21 S22 S23 S24 S25
FA 0 1 1 1 1 0 0 0 0 0
S11 0 0 0 0 0 1 1 0 0 0
S12 0 0 0 0 0 0 1 0 1 0
S13 0 0 0 0 0 0 0 1 0 0
S14 0 0 0 0 0 0 0 0 1 1
S21 0 0 0 0 0 0 0 0 0 0
S22 0 0 0 0 0 0 0 0 0 0
S23 0 0 0 0 0 0 0 0 0 0
S24 0 0 0 0 0 0 0 0 0 0
S25 0 0 0 0 0 0 0 0 0 0

Table 2 Distance matrix (km)

FA S11 S12 S13 S14 S21 S22 S23 S24 S25
FA 0 810 960 6570 7600 8830 1440 9170 9180 9390
S11 810 0 800 5170 8370 8350 1970 8580 9670 9800
S12 960 800 0 6180 8400 9120 1360 9290 8910 10,050
S13 6570 5170 6180 0 13,430 5040 7540 4300 13,760 13,510
S14 7600 8370 8400 13,430 0 11,360 7670 12,570 7800 2420
S21 8830 8350 9120 5040 11,360 0 10,260 1230 18,000 9890
S22 1440 1970 1360 7540 7670 10,260 0 10,550 7740 9320
S23 9170 8580 9290 4300 12,570 1230 10,550 0 18,020 11,110
S24 9180 9670 8910 13,760 7800 18,000 7740 18,020 0 9780
S25 9390 9800 10,050 13,510 2420 9890 9320 11,110 9780 0

assumed in the calculation that the material cost (cmi) is 5 Eur/pieces all over the
world. The specific transportation cost (cti) inside Europe is 0.00024 Eur/Km (road
transport), between Europe and Asia and between Europe and America is
0.00012 Eur/Km (possibility of water transport).
The objective function in the optimization is the total cost (Eq. 7), which
including production costs at the final assembler and at the suppliers and trans-
portation costs between the members (in our example the inventory cost, costs of
service providers, operation cost of the virtual enterprise and time consumptions of
activities are not taken into consideration).
Table 3 shows the most important data relating to the final assembler (FA),
primary suppliers (S1i) and secondary suppliers (S2i). Flexibility parameters and
production capacities are also defined in Table 3.
Table 3 Specific production- and material costs and flexibility parameters
cpi (Euro/piece) cmi (Euro/piece) Production Flexibility of the Flexibility of the IT Liquidity Organizational
capacity (pieces) manufacturing system infrastructure structure
FA 5 5 450 5 4 3 4
S11 5 5 250 3 3 3 3
Optimal Formation of Logistics Networks

S12 5 5 250 3 5 3 3
S13 2.8 5 250 3 3 5 3
S14 5 5 350 3 4 3 3
S21 2.8 5 200 3 3 3 3
S22 5 5 100 3 3 3 3
S23 2.8 5 250 3 3 3 3
S24 5 5 150 3 3 3 3
S25 5 5 350 3 3 3 3
399
400 G. Kovács et al.

3.2 Operation and Main Screens of the Software

Application

In the next part of the paper we show the most important print screens of the
developed software.
The software provides the possibility of determination of:
• data for the products to be produced,
• for data potential members of the supply chain and,
• relations for members of the supply chains (Fig. 3).
In the menu “Data for the products to be produced” we can define the parameters
of the product to be manufactured.
In the menu “Data for potential members of the supply chain” we can define the
most important parameters relating to the final assembler, suppliers and forwarding
service providers (Fig. 4).
In the menu “Relations for potential members of the supply chain” the relation
matrix, distance matrix and transport modes can be defined (Fig. 5).
In the menu “Results of the optimization” we can select the objective function of
the optimization (Fig. 6). In this paper we show the cost optimization procedure,
the time optimization and the multi objective optimization is under development.
Flexibility constraints relating to the chain members can be also set (Fig. 6).
Recently the optimization in case of single objective optimization is performed by
systematic search. This method is absolutely good in case of a small network, but
for a huge size network we will apply more robust optimization algorithm. The
multi objective optimization algorithm is also under development.
The result of the cost optimization is also can be seen in this screen. The possible
chain combinations—which fulfill the constraints—can be listed in the screen.
As it can be seen in Fig. 6, the optimal formation of the supply chain in our
example is FA–S13–S23, when the total specific cost of the supply chain will be
minimal.

Fig. 3 Main menus of the

program
Optimal Formation of Logistics Networks 401

Fig. 4 Screens for parameter setting of supply chain members

Fig. 5 Screens for parameter setting of relation matrix, distance matrix and transport modes
402 G. Kovács et al.

Fig. 6 Screens for selection of objective function, parameter setting of flexibility constraints and
results of the optimization

4 Conclusion

In this paper the definition and characteristics of supply chains and virtual enter-
prises were introduced.
Authors elaborated the objective functions and constraints for optimization of
virtual enterprises. Possible objective functions can be the minimization of total
Optimal Formation of Logistics Networks 403

cost and lead time of the chain, possible constraints can be also defined e.g. for the
production- and service capacities, inventories and flexibility of the chain members.
The research group developed a software application for formation of optimal
supply chains which was introduced in this article.

Acknowledgements This project has received funding from the European Union’s Horizon 2020
research and innovation program under grant agreement No. 691942. This research was partially
carried out in the framework of the Centre of Excellence of Mechatronics and Logistics at the
University of Miskolc.

References

1. Graham CS (1989) Integrating the supply chain., Int J Phys Distrib Mater Manag 19(8):3–8
2. Schönsleben P (2000) With agility and adequate partnership strategies towards effective
logistics networks. Comput Ind 42(1):33–42
3. Camarinha-Matos LM (2001) Execution system for distributed business processes in a virtual
enterprise. Future Gener Comput Syst 17:1009–1021
4. Gunasekaran A, Lai K, Edwin Cheng TC (2008) Responsive supply chain: a competitive
strategy in a networked economy. Omega 36(4):549–564 (2008)
5. Tamás P, Illés B (2015) The concept of a virtual logistics center for a Hungarian Region.
J Prod Eng 18(2):107–110
6. Esposito E, Evangelista P (2014) Investigating virtual enterprise models: literature review and
empirical findings. Int J Prod Econ 148:145–157
7. Skapinyecz R, Illés B (2014) Presenting a logistics oriented research project in the field of
E-marketplace integrated virtual enterprises. Appl Inf Sci Eng Technol Selected Topics Field
of Prod Inf Eng IT Manuf Theory Pract 197–211
8. Illés B, Buczkó K (2008) Virtual logistics network for supporting the supplier selection. In:
Proceedings 22th MicroCAD International Science Conference Miskolc (Hungary),
20.03.2008.–21.03.2008, 7–12
9. Telek P (2010) Operation strategies for delivery models of regional logistic networks. Adv
Logis Syst Theory Pract 4:33–40
10. Gubán M (2011) Non-linear programming model and solution method of ordering controlled
virtual assembly plants. Proc Logistics—the Eurasian bridge: materials of V. International
scientifically-practical, Krasnoyarsk (Russia), pp 49–58
11. Java Programming, https://en.wikibooks.org/wiki/Java_Programming
12. Simon K (2009) Object oriented programming using Java. ISBN 978-87-7681-501-1, 1st edn.
Simon Kendal&bookboon.com
 Part IV
Welding
Development of Complex Spot Welding
Technologies for Automotive DP Steels
with FEM Support

László Prém, Zoltán Bézi and András Balogh

Abstract In this paper spot-welding technology of sheets made of Dual-Phase

(DP) steel, which are one of the most important materials in the automotive
industry, has been analyzed with experimental research and finite element mod-
elling. Besides conventional mild steels that have been dominant for decades in
automobile production, the application of advanced high-strength steels (AHSS)
comes into focus. Among the first generation AHSS, DP steels are of the utmost
importance concerning their automotive use. In Hungary a lot of small and medium
sized enterprises function as the suppliers of the big automakers settled in our
country. Nowadays these enterprises, besides the welding of conventional mild
steel parts, often have the task of joining ferrite-martensitic DP steels with spot
welding. During the spot welding of DP steels, when using the technology familiar
with mild steels, the risk of hardening, the unfavourable failure of the joint or even
cracking during operation must be taken into account, therefore welding technology
must be planned on different theoretical bases. The traditional, continuous energy
input and the symmetric double pulse as the non-continuous energy input were
compared for resistance spot welding, focused on the advantages of pulsed energy
input. For numerical analysis an axisymmetric coupled finite element model
(FEM) is developed to study the effect of welding time and current intensity on
nugget size and thermal history in resistance spot welding process using MSC.Marc
software package. The cross-section macrostructures of the welded specimens are
examined and compared to the predicted size of the weld nugget and heat affected
zone (HAZ). Shear testing, cross tension testing, peel testing and hardness mea-

L. Prém (&)
A. Balogh

University of Miskolc, Miskolc, Hungary
e-mail: metpreml@uni-miskolc.hu
A. Balogh
e-mail: balogh.andras@uni-miskolc.hu
Z. Bézi
Bay Zoltán Nonprofit Ltd, Miskolc, Hungary
e-mail: zoltan.bezi@bayzoltan.hu

© Springer International Publishing AG 2017 407

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_36
408 L. Prém et al.

surements are also performed on specimens for each welding process to examine
the effect of the changes in different welding parameters on the load bearing
capacity of the joint.

1 Introduction

The automotive industry is one of the propulsive industries of Hungarian economy.

In Hungary several automotive clusters exist which represent hundreds of small and
medium sized enterprises which function as the suppliers of the big automakers
settled in our country. Thanks to the high-expenditure automotive material devel-
opments in the world, engineers working at these automotive suppliers today may
encounter steel grades with different strength and deformability during their work. It
follows that the processing of these materials with the various forming and welding
technologies can be very different, which could mean a great challenge in certain
cases for the engineers designing these technologies. For decades the low strength
unalloyed (so called mild) steels were decisive in the automotive industry because
of their good deformability and excellent weldability. Due to the need for
self-weight reduction of passenger cars, the ever increasing safety standards and
gradually tightening environmental specifications the application of conventional
and advanced high-strength steels came into view in recent years. Among the
first-generation advanced high-strength steels (AHSS) the DP steels with a
ferrite-martensite microstructure play a leading role concerning the application ratio
and their processing with spot welding [1]. However, DP steels possess prominently
high strength and significantly lower deformation ability compared to mild steels
which are otherwise excellent for spot welding. Therefore the engineers planning
the spot welding technology must take into consideration that the proven tech-
nology of the spot welding of mild steels might cause local hardness peaks,
hardening, decreasing of load bearing capacity, as well as an unfavourable failure
mode in the joints of DP steels [2].
The automotive sheets are mostly joined by some kind of resistance and arc
welding process, or in some special cases (e.g. in the case of poorly weldable or
non-weldable materials) by joining methods belonging to the welding related
technologies. As a result of innovations and developments in recent years, among
automotive joining technologies we can observe the application and slow expansion
of processes such as adhesive bonding, soldering and various mechanical methods
(riveting, clinching, etc.). However, it is an indisputable fact that today resistance
spot welding is still the leading joining method of automotive steel sheets.
Compared to the above listed joining methods resistance spot welding is the most
advantageous from engineering and economic point of view, so currently this is the
determinative method in car production and it can be safely predicted that it will
Development of Complex Spot Welding Technologies … 409

3% 1% 3% BH
4% 4% 10%
1% HSLA
DP
IF
MART
TRIP
CP
74% Others

Fig. 1 Application ratio of welding and other joining methods applied in the production of the
body in white unit of luxury cars [5]. Application ratio of different type of automotive steels in a
modern passenger car [6]

keep its dominance in the near future. It can be clearly seen in Fig. 1 as well that in
the production of modern, luxurious passenger cars the application ratio of spot
welding is approximately equal to the total ratio of all the other joining methods
applied. The plastic shaping technologies, the CAD/CAM systems, as well as the
continuous development and evolution of finite element software made it possible
for the car manufacturers and their suppliers to carry out more accurate strength
calculations and produce larger compound parts with a complex geometry.
Consequently the number of parts to be joined by welding and this way the number
of welding spots decreased. The foregoing is proved by the fact that while some
years ago an average passenger car’s integral body and frame had 4000…5000 spot
welds [3], this number was reduced to 2800…3500 [4] in the case of cars produced
today (eg. Audi, VW, BMW, Porsche).

2 Dual-Phase Steels in Automotive Industry

The development of DP (Dual Phase) steels were started in the early 1970s with the
aim of creating a type of steel with much better deformability within the strength
range of HSLA steels [7]. Since the main user was the automotive industry,
410 L. Prém et al.

measures characterizing cold-workability (such as strain hardening exponent, hole

expansion characteristics, minimum bending radius, spring-back etc.) gained sig-
nificant role besides strength and elongation after fracture.
The main reason for the development of DP steels was the rising of crude oil
prices due to the oil crisis caused by the 1973 Arab-Israeli war (Yom Kippur) and
the response measure taken to mitigate the effects of this crisis, namely to produce
passenger cars with a low fuel consumption. In favour of self-weight reduction
(from among many other possibilities) the reduction of wall thickness of steel parts
seemed to be the most efficient and have the most favourable cost implications, but
this required steels with higher strength. On the other hand reducing production
time and costs required such an upgrading of cold forming technology which
needed the approximation of the high deformability of deep-drawable mild steels
besides a twice-three times higher strength.
The expression ‘Dual-Phase steel’ was first used by Hayami and Furukawa
(Japan) to emphasise the heterogeneous microstructure compared to mild steels
thought to be Single-Phase (LSS) [8]. Despite its inaccuracy, this expression
quickly spread and became generally used.
DP steels contain fine, dispersedly distributed, hard martensite islands embedded
in a ferrite matrix. The soft ferrite ensures high deformability, whereas the hard
martensite increases the strength of steel. By controlling the chemical composition,
more precisely the carbon content and cooling rate, the ratio of martensite is
variable between 20…70% [9]. This heterogeneous microstructure is usually
achieved by continuous heat treatment (in a continuous annealing furnace) or by
controlled cooling after hot rolling [4]. The essence of these heat treatments is that
the desired ferrite/austenite structure is formed within the A1…A3 temperature
range, then the steel is cooled down at high rate from the intercritical temperature,
this way martensite is produced from the austenite during the process. The greatest
difference of the production of DP steels compared to the normal hardening of
steels is that the austenitization is only partial, therefore the austenite crystallites are
surrounded by ferrite [10]. DP steels possess very high tensile strength
(Rm = 600…1000 MPa), high toughness and relatively high deformability
(A80 = 10…18%), as well as an excellent energy absorbing capacity in the case of a
high-speed deformation. It follows from these very advantageous properties that the
bumpers, A, B, C pillars and various reinforcing elements of the body (doors, roof
etc.) of a passenger car could be primarily produced from the DP steel type and
through their significant energy absorbing capacity these elements are intended to
serve the safety of passengers during an incidental collision. Figure 2 demonstrates
the application possibilities of DP steels in a modern car’s body in white unit
manufactured today [11].
Development of Complex Spot Welding Technologies … 411

Fig. 2 Application of DP type Docol steels in a modern passenger car’s body in white unit [11]

3 Resistance Spot Welding of Automotive Dual-Phase

(DP) Steel Sheets

From among the available types of DP steels we have carried out experiments on
the DP 1000 material which possesses the highest strength and lowest deforma-
bility, being the most challenging from a welding engineering point of view. The
nominal thickness of the Docol DP 1000 steel sheets purchased from Swedish
manufacturer SSAB was 1.0 mm. During the examination of the spot weldability of
the experimental steel we examined the microstructure of the base materials
(ferrite-martensite ratio), we determined the mechanical properties with destructive
testing and the chemical composition of the sheets was checked with a spectrom-
eter. The results of the tests are summarized in Tables 1 and 2.
During welding it has to be taken into consideration at any case that the base
material (BM) contains so called ‘soft’ martensite with low carbon content, and that

Table 1 Mechanical properties of DP 1000 steel

Steel Mechanical properties
grade Rm Rp0.2 Rp0.2 / Rm A80 Ratio of Average
(MPa) (MPa) (−) (%) martensite hardness
(%) (HV0.2)
DP 1000 1047 721 0.69 11.33 55 324
412 L. Prém et al.

Table 2 The chemical composition of experimental DP 1000 steel given in mass percent
Steel Chemical composition
grade C Si Mn P S Nb V B(%)
(%) (%) (%) (%) (%) (%) (%)
DP 1000 0.148 0.49 1.50 0.010 0.002 0.015 0.01 0.0004

the microalloying elements can facilitate the hardening of the austenitized material
volumes [12].
It is less known among welding experts that resistance spot welding also has its
own weldability conditions just as fusion welding processes. It is an expectation for
spot welding that the spot welds with the specified spot diameter should be created
reproducibly, without cracks and with a load capacity characteristic of the base
material and the joint type. Weldability strongly differs from that of arc welding
processes due to the characteristics of spot welding (fast heating, small weld pool,
compressive stress and crystallization during the intense heat-removal caused by the
copper electrodes), belonging to pressure welding processes. The test criteria of spot
weldability are usually given as the maximum hardness of joints and the appearance
of unfavourable failure modes in welds determined by a qualification method.
During the examination of spot weldability the effect of the chemical compo-
sition of the base material is expressed by the carbon equivalent, similar to fusion
welding. Equation (1) shows such a carbon equivalent, which was introduced by
Japanese researchers for the weldability qualification of automotive AHSS steels.
The 0.24% boundary value marks the limit where the strength (as well as the ratio
of cross tension strength and tensile shear strength) of the joints begins to decrease
as a function of the increasing strength of the base material [13].

Si Mn
CERSW ¼ C þ þ þ 2
P þ 4
S  0:24% ð1Þ
30 20

If the value CERSW  0.24%, the spot welded joint will predictably suffer plug
failure, however, if CERSW > 0.24% the joint will show partial or interface failure,
or will not suffer plug failure at all [14]. We have calculated the carbon equivalents
of Docol DP 1000 from the chemical composition of experimental sheet metal with
the help of Eq. (1), the values are summarized in Table 2.
The carbon equivalents of DP 1000 steel sheet exceed the 0.24% boundary
value, the exact value is 0.27%. This means that in the case of using the conven-
tional continuous energy input spot welding technology for the joints of DP 1000
materials cracking and/or unfavourable fracture modes can appear, or even the
premature fatigue failure or brittle fracture of joints might occur during operation.
Therefore, the appliance of a complex working programme with non-continuous
energy input, where the cooling rate can be controlled, is favourable in the case of
spot welding DP steels. However, even with such a complex working programme,
the cooling rate is greater than the upper critical cooling rate so the microstructure
of the spot welded joint will be 100% martensitic.
Development of Complex Spot Welding Technologies … 413

Fs Fs
Continuous energy input Symmetric double pulse
Iw Iw
t w1 =t w2

tw t w1 t int t w2
t t

Fig. 3 The continuous energy input and the symmetric double pulse as the non-continuous energy
input during RSW

We compared the traditional, continuous energy input and the symmetric double
pulse as the non-continuous energy input during our experiments (Fig. 3). The
effect of the intermediate time between the first and second pulses on the hardness
distribution, susceptibility to hardening, load bearing capacity and failure modes of
high strength steel joints, in case of spot welding with symmetric double pulse
current, have been studied by experimental and numerical method. Based on the
results of preliminary spot welding experiments conducted on DP steels we man-
aged to define such parameter combinations (Table 3) which make it possible with
both energy input methods to create spot welds with the highest possible and
approximately equal tensile shear strength [12].
Then we prepared tensile shear for checking, cross tension, peel test specimens
with the parameters of Table 3, because we assumed that the tensile shear test is
less sensitive to technological variants than the other two test methods and it does
not provide accurate information on the quality of spot welds in high-strength
steels, primarily on their susceptibility to cracking.
The tests were performed on a sample containing 11 specimens. Design tensile
shear strength, cross-tension strength and peel strength values were determined at
95% confidence level by using Student’s t-test. Macroetching specimens and hardness
distributions of joints examined during the comparison analysis are shown in Fig. 4.
The microscopic examination and the hardness measurement of welds gave very
exciting results. We found that in the case of applying 14…15 cycles of interme-
diate time the outer part of the weld nugget starts to become recrystallized in a ring
shape. This ring increasingly widens from the outside towards the inside with the
increase of the intermediate time. The recrystallized ring is a fairly mild part of the
spot weld with a good deformability, its hardness is approximately equal to the
hardness of the base material, while the hardness of the non-recrystallized internal
core with a coarse dendritic structure stays around 475…500 HV. In Figs. 4 and 5
the recrystallized ring and the internal core can be well observed [15].
The design shear strength of joints welded by symmetric double pulse is lower
by 5…6%, than the same of joints welded by continuous energy input. This value
approximately equals the rate of the differences in spot diameters. Despite the
smaller spot diameters, more favourable failure modes and better load bearing
414 L. Prém et al.

Table 3 Welding parameters of resistance spot welding experiments

Nomination Continuous energy input Pulsed energy input
Number of pulses: 1 2
Intermediate time: tint (cycle) 0 15
Weld time: tw (cycle) 12 6
Weld current: Iw (kA) 5.75 6.35
Electrode pressure: Fe (bar) 3 3

Fig. 4 Hardness distribution and spot weld of the joints of DP 1000 steel examined in the
experiment, welded a with a continuous energy input, b with a symmetric double pulse.
Magnification: 16. Etching: Nital

Fig. 5 SEM pictures from the structure of nuggets a dendrite structure with continuous energy
input, b tempered ring with symmetric double pulse. Magnification: 1000

capacities of joints welded by non-continuous energy input were observed during

cross-tensile and peel tests (Figs. 6 and 7).
The results of the tension tests were proved that using an appropriately increased
intermediate time between pulses the outer part of weld nugget can be recrystallized
and tempered with the second current pulse, the microstructure of spot welds in DP
steels can be fined, their hardness distribution and the conventional static
mechanical properties (summarized in Table 4) and the failure mode of the joints
can be significantly improved [12, 15].
Development of Complex Spot Welding Technologies … 415

Fig. 6 Comparison the cross tension test results

Fig. 7 Comparison the peel test results

Table 4 Summary of the results of tensile shear, cross tension, and peel tests
Nomination Energy input
Continuous energy input Pulsed energy input
Tensile shear force: Fs (kN) 12.38 11.93
Cross tension force: Fc (kN) 3.44 4.08
Peel force: Fp (kN) 0.95 1.24
416 L. Prém et al.

In order to appreciate the effects of microstructure on the dynamic load capacity

and hardness distribution of the joints, it is important to recognize the various weld
regions. In the Fig. 4 the typical welding zones can be identified clearly. The fusion
zone (FZ) is created by heating above the melting point, while the surrounding heat
affected zone (HAZ) material itself consists of several regions which experience
thermal cycles with progressively decreasing peak temperature from the fusion
boundary. Outside the fusion zone, near the fusion boundary, the peak temperature
exceeds the Ac3 upon which 100% austenite forms. This region is called the
supercritical HAZ. Further from the fusion boundary, the peak temperature is within
the intercritical (Ac1 − Ac3) range, which results in austenitization of the
carbon-rich martensite phases while large islands of undissolved ferrite remain
unchanged. The cooling rate is typically high enough to result in the decomposition
of austenite to hard metastable phases such as martensite, bainite and carbides.
Thus, the room temperature microstructures of the fusion zone, the supercritical and
intercritical regions consist of a higher volume fraction of hard phases than the base
metal. In the subcritical HAZ, the peak temperature is below the Ac1 resulting in
tempering of metastable martensite. In general, the subcritical area will soften
during the welding of DP steels. The soft zone is partly attributed to the disap-
pearance of martensite and partly due to tempering of the remaining martensite
pre-existing in the base metals during RSW [16].
In order to analyse the effect of RSW processes on the microstructure of the spot
welds and the welding metallurgy of the DP 1000 steel numerical models must be
used. The microstructure of the welded joint depends heavily on the base metal
structure. So it is important to know the thermal history and the effects of the
different regions on the resulting microstructure.

4 Finite Element Modelling

FE modelling of the spot welding process can be difficult for most of the modelling
tools including finite element based software, as RSW is governed by
electrical-thermal, mechanical and metallurgical phenomena. It is difficult to sim-
ulate the RSW process because three different physical phenomena are interacting
with each other. The model takes the following physical and metallurgical inter-
actions into consideration in the simulations: interaction between the
electro-kinetics and heat transfer via the Joule effect, heat transfer and phase
transformations through latent heat and heat transfer, electro-kinetics, and
mechanical behaviour via contact conditions.
The welding process starts with analysing the squeeze cycle in which electrode
force is applied to the electrodes. The results of this mechanical analysis include
initial deformations and contact area, which serve in electro-thermal analysis. In
this stage, the temperature distribution by Joule heating is calculated for an incre-
ment from the fully coupled electrical thermal FEA. In the electrically thermally
coupled analysis the electrical and thermal boundary conditions are applied to the
Development of Complex Spot Welding Technologies … 417

model in a house. The calculations of Joule heating at the sheet-sheet and

electrode-sheet faying surfaces, as well as in the base material and electrode are
then performed. As a result, the temperature distributions are obtained in the first
increment and sent to the mechanical analysis as a nodal thermal load. Contact
pressure and deformations are the results of mechanical analysis that obtains a new
contact condition. So, the mechanical results are transferred to the electro-thermal
analysis to update contact conditions for the next increment analysis. This loop
continues until the welding time is finished.
Numerical calculations were performed with commercial finite element
(FE) software package MSC.Marc and Simufact.welding solver. The problem was
solved numerically in a cylindrical coordinate system due to axial symmetry of the
geometry. The actual geometry of the electrode was measured and replicated in the
model. The associated element mesh is shown in Fig. 8. The finite element mesh
contains 4095 linear quadrilateral elements and 3822 nodes.
Thermal-electric and mechanical boundary conditions were applied to the FE
model. Heat transfer to the surrounding air, using convection and 20 W/m2K
convective heat transfer coefficient was used. Temperature at the far end of the steel
sheet and the water temperature inside the copper-chromium-zirconium (Cu–Cr–Zr
alloy) electrodes were both assumed to be 20 °C. At the upper electrode, current
and mechanical pressure was applied uniformly to the top of the electrode.
The elastic behaviour is modelled using the isotropic Hooke’s rule with
temperature-dependent Young’s modulus. The thermal strain is considered using
thermal expansion coefficient. The yield criterion is the Von Mises yield surface. In
the model, the strain hardening is taken into account using the isotropic Hooke’s
law [12].

4.1 Material Properties

In order to capture the correct microstructure evolution, all of the relevant

mechanical, thermal and electrical properties of the sheets and electrodes must be

Fig. 8 Associated finite

element model
418 L. Prém et al.

Table 5 Electrical properties

Electrical Electrical E. contact E. contact
resistivity DP resistivity conductance conductance
1000 (Ωm*10−7) electrode (Ωm*10−7) sheet/sheet (1/Ωm*109) electrode/sheet
(1/Ωm*109)
20 2.432 0.257 0.203 1.075
100 2.975 0.315 0.298 2.174
200 3.756 0.397 0.401 3.334
300 4.656 0.487 0.505 4.348
500 6.985 0.675 0.714 6.668
700 10.03 0.845 0.909 8.621
1000 14.44 1.001 1.222 12.05
1300 15.87 − 1.538 16.67
1600 17.51 − 1.852 18.52
2000 17.62 − 3.025 26.13

known. Since electrical and physical properties vary with temperature and are not
readily available, many of these values were generated with JMatPro software and
assumed to be homogeneous. Contact resistivity of the sheet-to-sheet and
electrode-to-sheet interfaces were also assumed to vary with temperature (Table 5).
Material properties (i.e. thermal conductivity, specific heat, thermal expansion
coefficient, density, Young’s modulus and yield strength) for steel having nominal
tensile strength of 1000 MPa were considered to be temperature dependent.
Example of properties at room temperature has been listed in Tables 1 and 6.
For numerical simulation of RSW, the accuracy of critical austenite transfor-
mation temperatures is crucial. The start temperature for the austenite transforma-
tion depends strongly on three essential parameters, namely heating rate, initial
microstructure and chemical composition (Table 2). Ferrite-austenite transforma-
tions at high heating rates are characterized by JMatPro welding-cycle interface.
The temperature and strain rate depend flow curves are calculated by JMatPro
Simufact.premap interface with 30 µm initial grain size starting at 1300 °C for each
phase. A similar calculation has been done in [17].
Cu–Cr–Zr alloy, which has high thermal conductivity performance, is selected
for electrodes. Non-linear time dependency of thermal and electrical material
properties as well as convection coefficient rate for water, air and gas flow are all
obtained from literature and paper [17].

Table 6 Physical and mechanical properties at room temperature

Thermal Specific Thermal expansion Young Density
conductivity heat coefficient (1/K*10−6) modulus (kg/m3)
(W/mK) (J/kgK) (GPa)
DP 1000 55 448 12.8 218 7850
Cu–Cr– 450 380 18.5 129 8840
Zr alloy
Development of Complex Spot Welding Technologies … 419

5 Comparisons Between the Numerical Calculation

and Experiment Results

A simulation model has been developed and extensive numerical calculations were
carried out to find out the spot diameter and hardness distribution of resistance spot
welded DP 1000 joints. Figure 9 shows the thermal cycle curves in different
locations calculated by MSC.Marc software according to the welding parameters
listed in Table 3.

Fig. 9 Temperature history: a continuous energy input, b symmetric double pulse

420 L. Prém et al.

Fig. 10 Comparison of simulated peak temperature map to experimental micrograph

Predefined mesh nodes are identified in Fig. 9. The location of these nodes was
used to compare the thermal history within the sub-regions for different RSW
parameters. During the single-pulse RSW process (Fig. 9a), the DP steel was heated
to the peak temperature of 2530 K at an average rate of about 8000 K/s. After
reaching the peak temperature, the steel was quenched to about 300 K with
4500 K/s during where the martensitic transformation can occur. Figure 9b shows
the modelled temperature profile at the reference nodes for double-pulse RSW.
After the heating of the first pulse the specimen was cooled to about 550 K, where
partial austenite may transform into martensite, and then it was reheated by adding
the second pulse. The reheating effect of second pulse was predicted to occur in all
weld regions. For the inner-part of the FZ region the temperature range is typically
1200–1300 K at the reference nodes 1984–1991. The outer-part of the FZ and
intercritical (IC) region peak temperature was predicted (node 1994–1998) to be
1100 K and typical range from Ac1 and Ac3. The peak temperatures near the node
1994 was exceed Ac3, resulting in a fully austenitized local structure. The short time
above Ac3 was limited grain growth, producing an ultra-fine structure upon cooling
in this region.
Figure 10 compares the peak temperature contours with experimental
cross-section micrographs. The FZ, HAZ and BM region can be clearly observed.
The grey region mostly represents the FZ and the black grey region mostly rep-
resents the BM that is not affected by heat. The regions with other colours are the
HAZ, where the material is partially transformed into austenite. The size of FZ and
HAZ is again in good agreement with the experimental observations.
The microindentation hardness profiles (measured and simulated) of welded DP
1000 steel joints are shown in Fig. 11. Significantly higher hardness values,
approximately 1.5 times higher than that in the base metal were observed in FZ
zone. The profile was relatively flat across the fusion zone with an average FZ
hardness of 490HV. In the IC HAZ of joints near the fusion boundary also
exhibited high hardness, dramatically decreasing toward the unaffected base metal
Development of Complex Spot Welding Technologies … 421

Fig. 11 Comparison of calculated and measured hardness data a continuous energy input,
b symmetric double pulse

(324HV). A softened zone was observed in the outer-HAZ where the hardness was
significantly lower (230HV) than the base metal.
Figure 11b shows how the second pulse affects the measured hardness values.
For a second pulse softening of the FZ was a result of low cooling rates. This
suggests an increase in the transformation time and auto-tempering of the
martensitic structure during cooling [18]. The results of the numerical simulation
422 L. Prém et al.

support this experimental result. So as to fully understand softening kinetics and the
relationship between microstructure and mechanical properties a more extensive
study should be done on this type of DP steel with various welding parameters.

6 Conclusions

Concerning the experiments and numerical modelling of resistance spot welding of

automotive DP steel sheets, the following conclusions can be drawn:
1. The results of the tests show that the failure mode of joints, the plastic strain
reserves and the conventional static load bearing capacity can be improved by
applying symmetric double pulsed energy input. Because of its many advan-
tages it is more favourable to use pulsed energy input for the spot-welding of
such automotive parts and load bearing elements produced from advanced high
strength DP steels, that are important for life and property safety (e.g. B-pillars).
2. The simulation shows that the second pulse can improve the weld strength of the
welded joint. The results of the FE simulations can be used for in process
planning of RSW in modern manufacturing. The results show that the accuracy
between physical experiments and numerical simulations is comparable to the
variations of weld output in practical welding. Thus, the prediction accuracy of
simulations is useful for engineers using process planning.
MSC.Marc simulations can be used to predict post weld hardness, adding some
metallurgical capability to the software. Additional investigations are required to
determine which the most suitable combination of the welding parameters for the
achieving of the ideal hardness distribution.

Acknowledgements The research work presented in this paper based on the results achieved
within the TÁMOP-4.2.1.B-10/2/KONV-2010-0001 project and carried out as part of the
TÁMOP-4.2.2.A-11/1/KONV-2012-0029 project in the framework of the New Széchenyi Plan.
The realization of this project is supported by the European Union, and co-financed by the
European Social Fund.

References

1. H. World Steel Association (2009) Advanced high strength steel (AHSS) application
guidelines, Version 4.1, pp 1–16
2. Prém L (2014) Spot welding experiments of automotive Dual-Phase steel sheets. Publications
of the MultiScience. In: Proceedings 28th microCAD, international multidisciplinary
scientific conference, University of Miskolc (Hungary)
3. Weman K (2003) Welding process handbook. Woodhead Publishing Ltd and CRC Press LLC
4. Janota M, Neumann H (2008) Share of spot welding and other joining methods in automotive
production. Welding in the World
Development of Complex Spot Welding Technologies … 423

5. IIW White Paper (2012) Improving global quality of life through optimum use and innovation
of welding and joining technologies, p 36
6. Advances in high strength steels for automotive applications, www.wordautosteel.org
7. ASM Handbook (2005) Properties and selection: irons, steels, and high performance alloys,
10th edn., vol 1, p 697
8. Tsipouridis P (2006) Mechanical properties of dual phase steels, PhD dissertation, Technische
Universität, München (Germany)
9. Dziedzic M, Turczyn S (2010) Experimental and numerical investigation of strip rolling from
dual phase steel. Arch Civil Mech Eng 10(4):21–30
10. Tisza M (2015) Material and technological developments in sheet metal forming with special
regards to the needs of the automotive industry. Arch Mater Sci Eng 71(1):36–45
11. DOCOL advanced high strength steels for automotive industry, www.ssab.com
12. Bézi Z, Prém L, Balogh A (2016) Development of resistant spot welding technology for
automotive ferrite-martensitic dual-phase steels with joint application of finite element
modelling and experimental research. Adv Mater Res 1138:43–48
13. Oikawa H, Sakiyama T, Ishikawa T, Murayama G, Takahashi Y (2007) Resistance spot
weldability of high strength steel (HSS) sheets for automobiles. Nippon Steel Technical
Report, No. 95
14. SSAB: welding of AHSS/UHSS steel. A guide for the automotive industry
15. Prém L, Balogh A (2015) Symmetric double pulse with increased intermediate time for
resistance spot welding of ferrite-martensitic DP steel series. In: Young welding professionals
international conference, YPIC 2015, Budapest (Hungary)
16. Khan MI, Kuntz ML, Biro E, Zhou Y (2008) Microstructure and mechanical properties of
resistance spot welded advanced high strength steels. Mater Trans 49(7):1629–1637
17. Bézi Z, Baptiszta B, Szávai Sz (2014) Experimental and numerical analysis of resistance spot
welded joints on DP600 sheets. BID-ISIM welding and material testing 23(4):7–12
18. Khan I, Kuntz M, Zhou Y, Chan K (2007) Monitoring the effect of RSW pulsing on AHSS
using FEA (SORPAS) software. SAE Technical Paper 2007-01-1370
A Lightweight Design Approach
for Welded Railway Vehicle Structures
of Modern Passenger Coach

István Borhy and László Kovács

Abstract The design of railway vehicle structures is a complex task requiring

absolute care, well-founded theoretical knowledge and significant practical expe-
rience. The requirements related to the vehicle structures make it necessary to
manufacture structures that have the right amount of stiffness and are adequate also
in terms of fatigue resistance, but have the lowest weight possible and can eco-
nomically be manufactured, and at the same time are simple and easy to maintain.
The main aspects of designing for modern welded vehicle structures are, as follows:
suitable load-bearing capability and safety, the correct construction, according the
welding technology, and the same, the cost-effectiveness, as well. We can achieve
significant reduction of weight by increasing of the strength of elements and by
reduction of the thickness of plates and construction elements. The minimum of the
necessary thickness of plates can be calculated by optimization of sizes. In our
lecture we will present the weight reduction design approach, and new innovative
design methods and technologies for lightweight welded railway structures of the
modern passenger coach by the examples of new IC+type railway passenger couch
family, developed by MÁV-START Ltd.

1 Introduction

The development of railway transportations is the priority for the European Union
and for Hungarian National Transport Policy [1, 2]. The Hungarian government
specified the share incise and prove of competitiveness of railway transportation, as
part of National Railway Development Conception [3] in its government decision
about the Strategy of National Transportation Infrastructure Development,

I. Borhy (&)
TÜV Rheinland InterCert Kft, Budapest, Hungary
e-mail: iborhy@hu.tuv.com
L. Kovács
MÁV-START Vasúti Személyszállító Zrt, Budapest, Hungary
e-mail: kovacs.laszlo13@mav-start.hu

© Springer International Publishing AG 2017 425

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_37
426 I. Borhy and L. Kovács

in August 2014. To achieve these objectives, necessary to decry the average age of
rolling stock, it means refurbishment (main repair) or changing of weathered rail-
way fleet [4]. This could be a good possibility to revivify the traditional, domestic
manufacturing of railway stock, the same via innovative development it could
contribute the income generating capacity and the employment, improve domestic
small- and medium-sized enterprises in the field of sub-suppliers. As the first step of
this process, the MÁV-GÉPÉSZET Ltd. in 2012—on the base of EU support and
co-financing of European Regional Development Found—started to design the first
members of IC+type railway passenger couch family and the manufacturing of the
first example [5].
This lecture describes the designing, manufacturing and conformity assessment
of the body structure as a welded vehicle structure, via presentation of examples of
engineering, manufacturing and conformity assessment of 2nd class coach.

2 Presentation of IC+Type Railway Passenger Coach

Family

MÁV-GÉPÉSZET Ltd. in 2011 got owner’s authorization to start own manufac-

turing of railway vehicles, as the “spine” of national railway passenger transport, to
develop national inventions, domestic manufacturing potential and to keep and
develop domestic employment. The goal of this project to meet the high demands of
passengers to develop a vehicle type Z1, which is economically manufactured and
maintainable, and which is able to travel with 200 km/h speed in domestic and
international traffic. The main elements of the family:
• 2nd class saloon type coach;
• Shared 1st class and bistro coach;
• Multipurpose grow wagon, for person with reduced mobility with wheel-chair
and transportation of bicycles;
• Driving car with 2nd class compartment interior.
As the first phase of the project, the developing and preparing of manufacturing
of two pieces 2nd class grow wagon was done. Vehicles have an air-conditioned
passenger area (Figs. 1 and 2), closed-circuit vacuum toilet, GPS controlled,
audio-visual passenger information system. Vehicles have an energy supply at each
seats and internet connection via free Wi-Fi [6].
The European Parliament and Council of European Union defined rules of
Interoperability of the rail systems Directive [7] for international mobility of rail-
way vehicles. This Directive defines rules interoperability of safety designing,
manufacturing, operation and maintenance of elements, the same, the roles con-
formity assessment and licensing of these elements. The adoption of these
Directives of the domestic rule of law is done [8], so the licensing by the National
Safety Authority is done, the type approval is issued (the number of type approval is:
A Lightweight Design Approach for Welded Railway … 427

Fig. 1 The prototype of IC+2nd class coach (Photo: Sándor Czeglédi)

Fig. 2 Saloon type interior arrangement of the 2nd class coach (Photo: Sándor Czeglédi)

UVH/VF/320/9/2015 (1st June 2015)). The precondition of the licensing is the

approval of the full compliance of requirements by a Notified Body. The process of
conformity assessment is done by experts TÜV Rheinland InterCert Kft, in coop-
eration with Luxcontrol Nederland B.V. and TÜV Rheinland InterTraffic GmbH.,
belonging to the TÜV Rheinland Group, according requirements of SB+SD audits
defined by resolution No.: 2010/713/EU [9].
428 I. Borhy and L. Kovács

On the base of traffic experiences of prototype vehicles, the Hungarian

Government issued Government Decision No. 1167/2016 (IV.6.) [10] about sup-
port of 100 pieces IC+type passenger coach. The serial manufacturing starts in 2017
in Szolnok.

3 The Carbody Structures of Railway Vehicles

The main construction part railways vehicles is the carbody structure, it task is to
ensure the strength and stiffness against the outsider dynamical stresses, the same
the reliable protection of passengers and intern area [11]. The forming of the
carbody structure—as the load-bearing element, is done according the principle of
lightweight construction. This lightweight construction decreases the weight of the
coach, via this fact it has positive influence to operational costs, decreases the load
of landing gears, which is important in the case of vehicles, running over 160 km/h
speed [12].
The engineering of railway vehicles is a complicated task, it demands a high
degree of care, a solid theoretical knowledge. Engineers have to accept and evaluate
a lot of viewpoints of—each other often conflicting—requirements for the aim to
meet the requirements. To meet all requirements is a hard task, necessary to accept
requirements of strength, functionality, reliability, aesthetics, etc., necessary to
accept the requirements of manufacturability, controllability, and to minimize the
costs of full lifetime (LCC). The typical failure mode if welded structures of
vehicles—in addition of the corrosion—appearing and developing of cracks in
result of the exhausting stress, thus for these effects have to be accepted during the
full process of engineering, and make all necessary measures to avoid these.
Over these, due attention has to be paid to the development of bright and
spacious interiors, heat and sound deadening and to ensure the flow and smooth
internal temperature during the elaboration of engineering concept of carbody
structure.
The requirements of the vehicle structure with sufficient rigidity and the wearing
stress is also suitable, but at the same time the lowest possible weight, economical
to manufacture and easy respectively. The development to maintain a low cost
structure is required. The carbody structures of passenger coach are characterized
by rib stiffened hull plate that bears the chassis together with all the load acting on
the vehicle. The individual structural elements plating themselves participate in the
rigidity of the structure and design of the vehicle’s longitudinal axis parallel forces
uptake. In case of a differentiated method of construction for each strengthen ele-
ments and panels are made out separately and are compiled into an assembly unit.
For integrated moulded method of construction elements, beam-profiles, which at
the same time provide the structural and cladding functions. The development of
modern passenger rail vehicles uses both structures carbody structures construction
methods characterized by: IC+passenger coach carbody structure was constructed
on the basis of differentiated method of construction (Fig. 3).
A Lightweight Design Approach for Welded Railway … 429

Fig. 3 Cross-section of the carbody structure and the welded structure

Fig. 4 Welded side wall elements (Photo: István Borhy)

The railway structures of the high-powered building will be designed according

to the principle by which the pre-built components (chassis, side panels, front, roof,
etc.) installed in the cabinet assembler machine welded together (Figs. 4 and 5).
The main structural elements of the carbody structures:
430 I. Borhy and L. Kovács

Fig. 5 Welded carbody structure in the assembling machine (Photo: István Borhy)

• Corrugated floors covered chassis;

• Sidewalls;
• End walls;
• Roof.

4 Design of the Carbody Structures of IC+Type Railway

Passenger Coach

The car body structure of the IC+type coach is developed according requirements,
mentioned above. The full life-time of coach is 30 years. During engineering was
paid attention to increase the quality and reliability of vehicles structure, and the
same to decry manufacturing costs. The elements of family meet requirements via
unified solution of platform-conception (Figs. 6 and 7).
The 3D design of IC+coach done with Autodesk Inventor 2012 software, the
cooperation between engineering groups is supported with Autodesk Vault cen-
tralized controller. The requirements of strength are detailed in standard MSZ EN
12663-1:2010 [13]. The standard categorizes the several vehicles into several
groups, the same, the kind and volume of test loads are categorized into groups, too.
The carbody structure if IC+type coach belongs to the category P-I.
The results of engineering have to be proved (verificated and validated) by
calculations and measuring to order assessment. The finite element calculations are
made by pre- and post-processing MSC Nastran 2012 with MSC SimXpert 2012
(Fig. 8).
A Lightweight Design Approach for Welded Railway … 431

Fig. 6 Layout drawing of IC+2nd class coach

Fig. 7 Section view of carbody 3D model

The strength measurements are made by MÁV Ltd. Vasúti Mérnöki

és Mérésügyi Szolgáltató Központ (VMMSZK) in the III. quarter of 2012 (Fig. 9).
The calculation and measurements proved strength of carbody structures. The
necessary constructional changings are made on the base of measurements results.
432 I. Borhy and L. Kovács

Fig. 8 3D FE model of carbody (Stress distribution and deformation due to standard loads)

Fig. 9 Strength test of carbody (Photo: Sándor Czeglédi)

A Lightweight Design Approach for Welded Railway … 433

5 Developing of Innovative Methods and Joining

Technologies

When we prepared the serial production of IC+coaches, the secondary aim was to
elaborate and develop innovative methods of design and production technologies.
The first target was to elaborate and prove the ability to use during the engineering
of construction and technology of welded railway structures, a multiple objective
function optimization procedure.
The environmental awareness developed some changing in manufacturing of
railway vehicles at the late ’90. The main method to reduce the mass of vehicles is
to use materials with increased strength. The main material for carbody structure
manufacturing are the unalloyed and low alloyed steels, but in result of the material
development, the percentage of strength/mass of used materials is better and better,
which has a positive influence on the load-bearing capacity and mass of vehicles.
The use of strengthen materials and decreased weight, demand to use less
energy, so it decries the operational cost. The moving of vehicles is possible with
less powered drives, or in the case of similar performance of the drive, the
dynamical characteristics of vehicles can be higher (speed, acceleration). The use of
high strengthen steels is provided by stricter regulations, which allow less and less
exhaust emissions.
The more effective method reduced the mass and cost is to make thicker
the plates and beams. The result is the thick-walled welded body construction. The
engineering of these constructions demands to accept a lot of questions. The
shrinkage of welded connection makes permanent tensions and warps, thin plates
and shells reformate and needs to damp vibration, these structures are sensitive to
inhibit warping. To avoid these effects the thin plate walls should be wrapped. The
optimal thickness of plates should be calculated. The main target was to reduce or
minimize the local and global deformation of vehicle structures.
One of the preferred technologies of vehicle structure manufacturing is the
resistance spot welding, which is proved by a lot of facts, for example the consistent
quality, effective manufacturing, easy to automate, etc. The engineering, economic,
environmental factors, the permanent developing if technology (non-continuous
energy input, decried the specific heat input) ensure, that this technology will be a
decisive technology. Our target is to define optimal technological parameters
(welding sequence, parameters), to explore the sensitivity of manufacturing against
these parameters.

6 Optimization of Spot Welded Sidewalls

Our presentation is part of our earlier research activity, which was reported for you
[14–17]. The sidewalls—as welded structures—have to meet requirements of
strength and stiffness, and the same have to meet aesthetical requirements, and the
434 I. Borhy and L. Kovács

manufacturing has to be economical, too. The economic manufacturing in this case

means to find the minimal manufacturing cost; the target described by the next
function:

X
n
k ¼ kmaterial þ kfabrication ¼ kmaterial qV þ kfabrication Ti ð1Þ
i¼1

where: kmaterial and kfabrication are manufactural costs, q is the material density, V is
the volume of construction, n is the number of spot welding phases (incl. posi-
tioning of electrodes and factual realization of welding), Ti is the time required for
the ith welding phase.
The engineering of the vehicle carbody meets requirements of MSZ EN 15085-3
[18], the annex F details requirements for quality of spot welded junctions. The
characteristics of weld performance class (CP C1-CP D), and requirements of
surface strength are fixed too. During the calculation of sidewalls, the requirements
of MSZ EN 12663-1:2010 are accepted, the determining the value of the fatigue
limit stress is important too. The fatigue life than limiting condition can be used in a
vehicle related to the welded constructions of optimization of construction [19]
(Figs. 10, 11 and 12).

Fig. 10 Compilation of 3D sidewall model and its drawing

A Lightweight Design Approach for Welded Railway … 435

Fig. 11 Stress and deformation on sidewall near window

Fig. 12 Finite-element model, equivalent stresses and deformations in the case of the HP-type
specimen (Fi = 100 N)

7 Conclusions

In this lecture we demonstrated questions of manufacturing, conformity assessment

of 2nd class IC+type passenger coach, developed by MÁV-START Ltd. We turned
the attention to the preferences of resistance spot welded constructions as a tech-
nological possibility in the in serial production. The actuality of this topic is proved
by the fact, the results of research are close to use in the serial manufacturing. Given
that the family development allows for further tests, so the development of applied
436 I. Borhy and L. Kovács

research with our work in the welded plate structures designing and manufacturing
optimization methods we intend to contribute.

Acknowledgements This presentation was conducted as part of the Project IC-PR01-GOP131/A,

on the basis of contract No.: K2378-11, between MÁV-GÉPÉSZET Ltd. and TÜV Rheinland
InterCert GmbH. The realization of this project is supported by the European Union, and
co-financed by the European Regional Development Fund.

References

1. White paper: roadmap to a single european transport area—towards a competitive and

resource efficient transport system, http://ec.europa.eu/transport/themes/strategies/doc/2011_
white_paper/white_paper_com(2011)_144_en.pdf, COM(2011) 144, Európai Bizottság,
Brüsszel, 2011
2. Government Decision: 1486/2014 (VIII.28.) Korm. határozat: National Transport Strategy
http://www.3k.gov.hu/remos_downloads/NTS_Strategic%20document_version%20for%
20public%20discussion_EN_2013.70.pdf, Hungarian Transport Administration, Budapest,
2013
3. National Transport Strategy (2013) National railway development concept. Hungarian
Transport Administration, Budapest
4. The business strategy of MÁV-START Ltd. (period from 2014 to 2023) for improve the level
of public transport (in Hungarian), Founders Decision: 10/280/2014 (20 Mar 2014)
5. Kovács K (ed) (2011) The MÁV-GÉPÉSZET Ltd. started the development of new railway
passenger coaches (in Hungarian), Railway Engineering (ISSN 0230-0605), No. 3, p 11
6. Mezei I (ed) (2014) Design and development of IC+type domestic InterCity passenger
coaches (in Hungarian) (ISBN 978-963-9852-20-4). MÁV-START Zrt, Budapest
7. Directive 2008/57/EC of the European Parliament and of the Council on the interoperability of
the rail system within the Community http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=
OJ:L:2008:191:0001:0045:EN:PDF, Official Journal of the European Union (17 June 2008)
8. Regulation of Ministry of National Development: 31/2010 (XII.23.) NFM rendelet: Process of
authorisation for placing in service, periodic inspection and the national vehicle registers of
the railway vehicles (in Hungarian) http://net.jogtar.hu/jr/gen/hjegy_doc.cgi?docid=
A1000031.NFM, Hungarian Official Journal, No. 197, pp 30412–30500
9. Commission Decision 2010/713/EC on the modules for the procedures for assessment of
conformity, suitability for use and EC verification to be used in the technical specifications for
interoperability adopted under Directive 2008/57/EC http://eur-lex.europa.eu/legal-content/
EN/TXT/PDF/?uri=CELEX:32010D0713&from=EN, Official Journal of the European Union
(9 November 2010)
10. Government Decision (2016) 1167/2016 (IV.6.) Korm. határozat: Development of the rail
vehicle industry in Szolnok and financial support of the production of IC+type passenger
coach (in Hungarian). Hungarian Off J 47:4295
11. Zobory I (ed) (2006) Handbook of railway engineering (in Hungarian) (ISBN 963 204 127
5). Magyar Államvasutak Zrt, Budapest
12. Radnai G (1976) Lightweight constructions in the vehicle and machinery industry (in
Hungarian) (ISBN 963 17 0765 2). Tankönyvkiadó, Budapest
13. MSZ EN 12663-1:2010—Railway applications. Structural requirements of railway vehicle
bodies. Part 1: Locomotives and passenger rolling stock
14. Borhy I, Schwartz I (2002) Practical application of FEM in the technology design of
resistance spot welding in railway production. In: Proceedings 3rd conference on mechanical
engineering (ISBN 963 699 1650), Budapest, (Hungary), pp 479–484
A Lightweight Design Approach for Welded Railway … 437

15. Borhy I, Szabó P (2005) Topical issues in the development of expert systems for use in the
process planning of resistance spot welded railway vehicle structures. Math Modell
Weld Phenomena 7 (ISBN 3-901351-99-X), Verlag der Technischen Universität Graz,
pp 1099–1109
16. Borhy I, Belső L (2012) Design and optimization of spot welded railway vehicle structures (in
Hungarian), Proc. 26th Hegesztési Konferencia (ISBN 978-615-5018-28-2), Óbudai Egyetem
(Hungary), pp 91–96
17. Borhy I, Belső L (2014) Design and conformity assessment of welded railway vehicle
structures, demonstrated via IC+project (in Hungarian). In: Proceedings 27th Hegesztési
Konferencia (ISBN 978-963-08-8585-0), Óbudai Egyetem (Hungary), pp 49–58
18. MSZ EN 15085-3:2008—railway applications. Welding of railway vehicles and components.
Part 3: Design requirements
19. Borhy I, Szabó P (2008) Possibilities of predicting the fatigue life of resistance spot welded
joints. In: Proceedings international conference on design, fabrication and economy of welded
structures (ISBN 978-1-904275-28-2), Horwood Publishing, Chichester (UK), pp 201–210
Challenges and Solutions in Resistance
Welding of Aluminium Alloys—Dealing
with Non Predictable Conditions

Jörg Eggers, Ralf Bothfeld and Thomas Jansen

Abstract For welding of aluminium alloys with Harms and Wende medium fre-
quency inverters (1000 Hz) we developed a special control mode AMC
(Aluminium Mode Classic) and its extension AMF to handle aluminium alloys
(mainly 5000th and 6000th alloys). These modes have been made to handle the
alloy groups. The new methods cover every day issues as undefined oxide layers or
other coatings. From the commercial side a standard inverter is used to reduce
investments. Internal monitoring while welding is another important topic of this
mode as well as constant weld time. The last item is the key topic for high volume
production customers. Examples are shown for several surface conditions. Methods
are suggested to preserve against crackles and cracks. In advance first tendencies in
quality monitoring are presented using the force signal during the current phase of
the welding process. In this paper we give the solution first before showing the
background.

Keywords Resistance welding
Adaptive control
MFDC inverter
Aluminium

alloys

1 Introduction

When dealing with aluminium joints of the system has to handle conditions
between the tips which are not known. Go one step ahead these conditions vary
from weld spot to weld spot.

J. Eggers (&)
R. Bothfeld
T. Jansen

Harms & Wende GmbH & Co KG, Hamburg, Germany
e-mail: joerg.eggers@harms-wende.de
R. Bothfeld
e-mail: ralf.bothfeld@harms-wende.de
T. Jansen
e-mail: thomas.jansen@harms-wende.de

© Springer International Publishing AG 2017 439

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_38
440 J. Eggers et al.

Fig. 1 Resistance drop, as R

described the first time in [1]

On the following pages solutions re provided which run out in the field and have
shown that such work stable.

1.1 Pre-conditioning

In the first step the uncertain situation has to be removed. When oxide layers are on
the surface of the material this causes a higher resistance as the base material, the
aluminium itself. Such coatings can be thicker or thinner depending how long the
material has been sitting on the shelf and how it was treated before reaching the
factory.
Hence what has to be done is to break through this layer to touch the base
material. Since the base parameter of the base material does not chance the first
phase is called a pre-conditioning phase. Figure 1 shows the measurable resistance
drop.

1.2 Force-Profile

A force profile can be used in the AMC case. A high electrode force is used during
the pre-conditioning phase and a lower during the weld and a higher force again
during the post welding phase. The lowered force results in a lower energy demand
due to the hither contact resistance.
A type of advanced process control suggested by DVS Merkblatt [2] (Fig. 2) and
before by [3].

Fig. 2 Suggested force and current profile from [2]

Challenges and Solutions in Resistance Welding … 441

2 Aluminium Mode Classic AMC

The aluminium mode classic combines the pre-conditioning phase and weld phase
and if needed the force profile.
Inside the inverter the weld is divided into the two sections as shown in Fig. 3.
Since the AMC is intended to use, for example, in automotive business the cycle
time becomes very important. One parameter to stop the weld process is, when the
pre-conditioning phase is not finished to a certain time segment. Then it is assumed
the conditions cannot be stabilized. The following weld phase is a constant weld
time segment. The sum of the maximum allowed conditioning phase plus weld time
is the maximum cycle time for the weld itself. Other times, as pre-hold time and so
forth have to be added.

2.1 User Interface XPegasus

The whole process control is assembled for the user in one XPegasus page. The
challenge in the design was to make the user feel familiar with a standard weld
schedule to keep the learning phase short.
This user interface is independent from the tools (Servo gun, pneumatic gun, or
7th Axis robot system). The user interface as in Fig. 4 is designed in the way that is
reminds to a normal Constant Current Regulated (CCR) weld. A graphical window
shows the entered process.

Fig. 3 Schematic diagram of AMC (Alu Mode Classic)

442 J. Eggers et al.

Fig. 4 XPegasus page for aluminium process control

As difference from the standard constant current schedule the break-through

resistance can be entered and the force profile can be programmed.

2.2 Weld System Set-Up

Figure 5 shows the gun the trials in Sect. 2.2 have been made with. It’s a servo
driven C-gun. This gun can keep up a pressure of 4000 N (899.23 lbf ). There is a
second gun in the lab with a larger transformer and a maximum force of 7000 N
(1573.66 lbf ). The inverter used in this system is a GeniusHWI424 (1600 A max
output current) with AMC installed. For power requirements, please refer to Sect. 5
later in this paper.
Almost flat D16 electrodes in standard alloy CoCrMo have been used. It’s
important to notice that no special electrodes have been used. These electrodes
show a large radius on both gun arms. However, one flat electrode can be used
single sided for better surface quality.
Challenges and Solutions in Resistance Welding … 443

Fig. 5 Servo-electric C-gun

2.3 Development of Process Factors

Using such a device you end up with following process diagrams:

Figure 6 shows a typical aluminium weld process using the AMC. The cure
moment of the process is the time frame between 900 and 1300 ms. The pink curve
shows the force profile. The green curve shows the resistance and the blue the
current.
When the current is applied at 900 ms the resistance jumps to a high level due to
the oxide layers and other resistance between the tips. After a short moment the
resistance curve drops. This is the moment when the tips break through the oxide
and touch the real aluminium.

Fig. 6 Process diagram resulted for AMC usage (XPegasus)

444 J. Eggers et al.

Direct after the current starts to raise the force value is moving slightly up and
then down again just left to the vertical line. This is the nugget build-up or in other
words the material expansion.
In this moment the inverter raises the current to the welding level of 23 kA.
For this weld we used a gun, which is usually intended for steel welds. The
reason is to show the weld process curves very clearly. Usually the gun would be
much stiffer and the drop in the force will be much less.
The force measurement shows a significant force drop when the material starts to
soften and a quite steep peak value when the nugget starts to grow. This peak is
related to the expansion material factor. Aluminium expands twice as fast as steel
(factor 24–12 on steel).

3 Pre-conditioning

The phase of pre-conditioning is combined by the force increase, followed by a

fading phase to the welding force. Then the welding force is reached the control
starts with a pre-conditioning current. In Fig. 7 is the case shown, that the resistance
threshold is reached relative lately.

3.1 Example with High Resistance Sn Coating

One example is shown here with high resistance Sn coating. Normally with the
standard welding method for aluminium (one pulse with no pre-conditioning) you
will get an expulsion because of the high start resistance.
With pre-conditioning this effect is reduced.

Fig. 7 Example for a long conditioning time of an old charge (red circle)
Challenges and Solutions in Resistance Welding … 445

Fig. 8 Example with high resistance Sn coating

When material is tin coated the initial resistance is higher (see circle in Fig. 8) as
on uncoated material. This increases the splatter risk by 20% especially on short
term welds.

4 Typical Welding Quality Until Tip Dressing

There is no hope to reach higher values at 20–50 welds. The main effect is electrode
pick up of aluminium. There every millisecond with high current power reduces the
lifetime.
The hardest effect is on the anode electrode (Fig. 9).

Fig. 9 Electrode erosion after 50 welds with different weld times of the high current main pulse [4]
446 J. Eggers et al.

Fig. 10 Imperfections
increased excessive the limits
of tip dressing cycles [4]

If you try to increase the number of welds between tip dressing you will get an
increase of surface cracks, strongly erode of surfaces and at the end the electrode
sticks on the aluminium plates (Fig. 10).

4.1 Methods to Decrease the Number of Imperfections

The main method is the usage of a longer downslope in current after the main pulse
combined with an increase in the electrode force. This will control the cooling
down of the spot. The volumetric change with time of the spot is reduced.
A typical control profile looks like Fig. 11.
The result is a nice looking spot like in Fig. 12.
Challenges and Solutions in Resistance Welding … 447

Fig. 11 Control profile to reduce the number of imperfections

Fig. 12 A spot without

imperfections maintained by
control profile (etchant side
view)

5 Price of High Currents and Welding Times

An important factor of a welding system is the power requirements. The most

common inverter in spot welding is the 80 kVA (max. 600 A output current) which
is widely used for steel welds of any kind.
In case of aluminium welds the inverter power class differs. Typical inverter
power classes used for such welds are 1600 A (i.e. GeniusHWI 2824) or 2400 A
(i. e. GeniusHWI2836) inverters. This has an impact not only on the fuse and on the
supply lines as well as on the power consumption of a plant. Table 1 below gives an
overview about technical key data of different inverter classes.
448 J. Eggers et al.

Table 1 Different inverter classes

Type HWI 2813 L 2816 L 2816 L 2816 W 2824 W 2836 W
Input
Nom. V 400–480, 400–480, 400–480, 400–480, 400–480, 400–480,
voltage 690a 690 690 690 690 690
Nom. current A 66 153 74 229 305 430
100% DC
Phases 3, PE, (N) 3, PE, (N) 3, PE, (N) 3, PE, (N) 3, PE, (N) 3, PE, (N)
Frequency Hz 50/60 50/60 50/60 50/60 50/60 50/60
Output
Voltage @ V 500 500 500 500 500 500
400 V
Voltage @ V 600 600 600 600 600 600
480 V
Nom. power kVA 130 300 145 450 600 845
@ 20% DC
Max. current A 800 800 1200 1200 1600 1600
Cooling qty. l/min Air 4 Air 4 4 4
Weight kg (lb) 21 (46) 21 (46) 21 (46) 21 (46) 24 (53) 24 (53)
a
690 V mains supply for certain volkswagen plants only

Hence, when a production is designed with a larger number of aluminium weld

cells, the infrastructure must be considered as well. An intermediate way is to limit
the load of the power system by a power divider or similar. This solution will have
an impact on the cycle time since the cells have to wait when the maximum load
allowed is reached.
Another factor on aluminium weld systems is the cooling. While cells welding
steel materials are typically equipped with air cooled inverters require aluminium
weld cells water cooled inverters. Hence, apart from the power consideration and
water network must be considered as well.
A third subject is the nominal voltage a system is operating in. In Europe, it’s
generally 400 V and in the US its 480 V. The difference in the voltage implies that
480 V systems have 20% mode energy as 400 V systems. This fact becomes very
important when designing a system.
Since power is calculated as P ¼ U
I the current flow in 480 V systems is lower
as in 400 V systems. This becomes even more important when systems are for
example designed in Europe for operation in the USA or the other direction.
Systems running well under 480 V may fail below 400 V since 20% of the energy is
lost and the system reaches its maximum.
Challenges and Solutions in Resistance Welding … 449

6 Extension with Quality Monitoring (AMF)

With an ideal electric motor servo gun it is very easy to construct a quality mon-
itoring system. You can simply take a tolerance band around a force reference curve
and from there you get a monitoring signal during current time.
This can be easily managed by several third party systems. In this chapter we
want to show the case for an older ‘used-up’ gun with a non-ideal force signal. It
will also show tendencies in future monitoring systems to build up robust
algorithms.
The force examples shown here are for non-ideal welding equipment. Especially
equipment with lower capabilities beyond 4 kN. This welding gun here has a larger
amount of lateral movement during force build up.
Here 1 mm  1 mm 6000 aluminium alloy plates are welded. We used 16 mm
electrodes of type A.
With a current of 30 kA we get a spot diameter of around 7 mm.
In Fig. 13 we see an increase in force of around 25–30 % during main current.
But the force signal is distorted by the lateral movement of the electrodes by larger
oscillations on the signal. The movement is not smooth and the mechanical system
is oscillating.
If we produce an expulsion with a higher current we see it in the force signal
clearly (Fig. 13). In the resistance signal no occurrence is observed.
One future requirement for expulsion detection (red circle in Fig. 14) is the
distinction between expulsion and lateral movement. The other requirement is the
measurement of the total amplitude by a robust filter mechanism.
Figure 15 shows a weld with a 2 mm spot diameter. So simple by signal
amplitude a differentiation might be possible. Here we reach only 10–15% increase
of the force during main weld current.

Fig. 13 Welding is OK. Reached 7 mm spot diameter

450 J. Eggers et al.

Fig. 14 Welding with expulsion in the late phase of the weld

Fig. 15 Welding for a not ok (red circle) welding with 2 mm spot diameter

With a clear force signal it might be no problem to get a monitoring signal. But
most times in practice it will not be the case because of the costs of the gun or
measuring equipment. So we have to find robust filter algorithms in future.

7 Summary

The Aluminium Mode Classic (AMC) and AMF are methods to join aluminium
materials. Typical materials used in automotive business are 5000 and 6000 series.
It bridges the problem with the undefined oxide layers on any aluminium. Such
layers caused problems with varying spot diameters. Stabilization of the weld
process is very important.
Aluminium has for natural reasons oxide layers on the surface the AMC and
AMF divide the weld process into sections:
Challenges and Solutions in Resistance Welding … 451

• Conditioning phase
• Weld phase
During the conditioning phase with a lower current i.e. 10 kA the inverter
monitors the resistance. When the resistance drops, it is assumed the aluminium is
touched by the electrodes and the weld begins with me. i.e. 35 kA.
The AMF extends the joining process with a monitoring function for quality
assurance.
There are some parameters to influence the weld, but important for the user is the
maximum weld time is fixed. Typical weld times for aluminium are in the range of
100–120 ms plus conditioning phase.
However a method is only as good as the entire system consisting of gun,
inverter and power supply. The examples in this paper showed the effect with too
weak gun arms. However, in this case such gun has been used to show the effects.
New power demands come up with welding aluminium materials. More pow-
erful inverters are needed.

References

1. Krevet A (2007) Untersuchungen zu Prozesseinflüssen beim Punktschweißkleben von

Aluminiumwerkstoffen unter Verwendung einer widerstandsgesteuerten Prozessregelung. Die
Verbindungsspezialisten, Tagungsband: Große Schweißtechnische Tagung, pp 638–639
2. Deutscher Verband für Schweißen und verwandte Verfahren e.V (1985) Widerstandspunkt-
und –Rollnahtschweißen von Aluminium und Aluminiumlegierungen von 0.35 bis 3,5 mm
Einzeldicke, Vorbereitung und Durchführung des Schweißers
3. Singh, S (1977) Beitrag zur Verbesserung und Sicherung des Tragverhaltens von
Widerstandspunktschweißverbindungen an Aluminium- und Stahlwerkstoffen durch technol-
ogische Maßnahmen und durch Entwicklung einer Regelungseinrichtung, Dissertation RWTH
Aachen (Germany)
4. NN Verbesserung der Prozesssicherheit des Punktschweißklebens von Aluminiumwerkstoffen
und Ermittlung von Verbindungskennwerten für Konstruktion und Simulation, Aschlussbericht
AiF 16.335N, GSI SLV, Duisburg (Germany), 2010
High Cycle Fatigue Investigations
on High Strength Steels and Their
GMA Welded Joints

Ádám Dobosy, János Lukács and Marcell Gáspár

Abstract High cycle fatigue tests were performed on two strength categories of
high strength steels, on quenched and tempered (S690 and S960) and thermome-
chanical (S960) types, on base materials and their gas metal arc welded joints, and
on different matching conditions. The planning and optimization of welding tech-
nologies based on investigations under cyclic loading conditions were built upon a
large number of investigations and statistical evaluation of the test results.
Statistical approach was already applied during the preparation of the investiga-
tions, which have been allowed the expansion of the scope of the results and the
increasing of their reliability. The article demonstrates and evaluates the results
comparing with each other and with literary data.

1 Introduction

In the automotive industry, as a dominant industry, one of the fundamental trends is

the weight reduction and the environmental load reduction, which can be achieved
with the application of high strength materials (steels [1, 2], aluminium alloys [3, 4]
and composites [5, 6]). In the case of the various vehicles, and vehicle structures the
dominant joining technology is the welding, with both the traditional and the
modern processes, and with the fusion and the pressure welding processes. Failure
statistics have long shown, that the third of the failures caused by fractures, where
cracks initiated in welded structures, furthermore the four fifth of the failures
occurred under cyclic loading condition [7].

Á. Dobosy (&)
J. Lukács
M. Gáspár

Institute of Materials Science and Technology, University of Miskolc, Miskolc, Hungary
e-mail: metda@uni-miskolc.hu
J. Lukács
e-mail: janos.lukacs@uni-miskolc.hu
M. Gáspár
e-mail: gasparm@uni-miskolc.hu

© Springer International Publishing AG 2017 453

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_39
454 Á. Dobosy et al.

During the welding processes, the joining parts were affected by heat and force,
which cause inhomogeneous welded joints. Because of these, zones with different
microstructure and different properties can be developed, and additionally stress
concentrator places can be formed. These changes can be caused faults, which
define the properties of the welded joint. In the case of cyclic loading condition the
failures have a more important role; however in the case of the vehicles the cyclic
loading is very conventional.
All these together justify, that our investigations were focused on the high
strength steels, from the joining methods on the welding technologies, and from the
loading conditions on the cyclic loadings. Three types of high strength steels
(S690QL, S960QL and S960TM), one fusion welding technology (shielded metal
arc welding, ISO 4063: 135), and high cycle fatigue (HCF) loading condition were
chosen for our investigations. In accordance with the welding challenges in our
days, the matching problem of the base material and the welding consumable was
examined as well. One of the main aims of our investigations is the determination
of high cycle fatigue design or limit curves for the tested steels and their welded
joints.
The wide range of our research work can be found in a book [8]; the details and
the results of the low cycle fatigue (LCF) and the fatigue crack growth
(FCG) investigations were summarized in [9, 10], and in [11], respectively.

2 High Cycle Fatigue Limit Curves

The wide variety of machines, equipment and structures are usually designed for
long operation, often running for several decades under cyclic loading condition. In
the case of the automobile structures, the number of cycles during the whole
lifetime could be millions and tens of millions. We talk about high cycle fatigue
when the loading is relatively low, and the number of cycles is relatively high,
between 104 (5
104) cycles and 108 (109) cycles. The cyclic loading of the
components, structural elements and structures can be very different, from the
simple mechanical stresses (tensile, tensile-compression, bending, torsion) to the
complex stresses (for example tensile and bending), furthermore, cyclic thermal
and/or environmental stresses can also occur.
The measurement results could be affected significantly by a number of factors.
This clearly means that under the same testing conditions, more test specimens must
be used, because of the relative high standard deviation of the test results.
Furthermore, mathematical statistical methods must be applied to the evaluation of
the results [12–14]. In the case, when test specimens from welded joints or whole
welded joints were examined, the effects of the welding process increase the
uncertainty further. The statistical approach is desirable in the preparation phase of
the experiments, so the validity range of the results can be widened and the reli-
ability can be increased as well.
High Cycle Fatigue Investigations on High Strength Steels … 455

In the area of the high cycle fatigue, one of the main research directions is the
determination of the design or limit curves. The high cycle fatigue limit curves can
be found in numerous prescriptions, and there are numerous known and applied
methods for the derivation of these curves, based on data from different fatigue and
non-fatigue examinations. A few, widely used regulations and design curves are as
follows: Eurocode 3 [15]; BS 7608 [16]; BS 7910 [17]; AASHO direction [18];
design curves based on empirical correlations [19].
The design or limit curves, found in the different specifications, were determined
based on experiments with different welded structures. The Basquin equation,
which describes the life-time region of these curves, is the following:

Nt ðDrÞm ¼ a; ð1Þ

where m and a are constants, depending on the materials and the conditions.

3 Welding Experiments on High Strength Steels

3.1 Material Properties and Weldability

For our studies, high strength steels from 690 and 960 categories were chosen,
welded with shielded metal arc welding (GMAW, ISO 4063: 135) process. In the
690 category quenched and tempered (Q + T) steels (S690QL and Weldox 700E),
and in the 960 category quenched and tempered (S960QL) and thermo-
mechanically (TM) rolled (S960TM) steels were chosen (EN ISO 10025-6). For
the welding experiments different filler materials were selected. The chemical
composition of the base and the filler materials and the mechanical properties can
be seen in Tables 1 and 2, respectively.
These steels have a higher energy after the production, than in the equilibrium
state, so the obtained microstructure can be irreversible altered during the welding
process. The heat affected zone (HAZ) can be easily hardened, furthermore, in case
of too large heat input the heat affected zone can be softened—because of the
coarsen of the microstructure—compared to the base material, which can cause
strength and hardness decreasing. It is a matter of course that both cases are
intolerable. Because of these effects, additional microalloys are used on these steels,
like aluminium, niobium, vanadium and titanium. Naturally, these features change
also affects the fatigue properties; the different quality of the different zones in the
welded joint (especially in the HAZ) significantly changes the fatigue resistance.
Besides all these, further undesirable phenomena can appear, e.g. different types
of cracks. Primarily, to avoid the cold cracking the workpiece must be preheated
before the welding, and it is necessary to limit the linear energy (Ev) during the
welding [20].
The complex task of the weldability of the high strength steels summarizes the
Graville diagram (Fig. 1) based on the carbon content and the carbon
 456

Table 1 The chemical composition of the examined base and filler materials, wt%
Material designation C Si Mn Cr Mo Ni S P Ti V Al
S690QL 0.14 0.30 0.96 0.60 0.19 – 0.002 0.009 0.02 0.005 0.05
Weldox 700E 0.14 0.30 1.13 0.30 0.167 – 0.001 0.007 0.009 0.01 0.03
S960QLa 0.16 0.23 1.25 0.2 – 0.04 0.001 0.008 0.004 0.04 0.06
S960TM 0.09 0.32 1.63 0.59 0.29 0.03 0.001 0.009 0.016 – 0.041
INEFIL NiMoCrb 0.08 0.50 1.60 0.30 0.25 1.50 0.007 0.007 – 0.09 –
Thyssen UNION 85c 0.07 0.68 0.61 0.29 0.61 1.73 0.010 0.006 0.08 0.01 0.01
Thyssen UNION X90 0.1 0.8 1.8 0.35 0.6 2.3 – – – – –
d
ESAB OK Tubrod 14.03 0.08 0.51 1.61 0.02 0.55 2.27 – – – 0.01 –
Thyssen UNION X96 0.12 0.80 1.90 0.45 0.55 2.35 – – – – –
a
Cu = 0.01, Nb = 0.016, B = 0.001, N = 0.003
b
Cu = 0.12
c
Cu = 0.06
d
Cu = 0.02
Á. Dobosy et al.
High Cycle Fatigue Investigations on High Strength Steels … 457

Table 2 The mechanical properties of the examined base and filler materials
Material designation Rp0.2, MPa Rm, MPa A5, % KV (−40 °C), J
S690QL 783 826 19 54
Weldox 700E 791 836 17 165
S960QL 1030 1076 16 56
S960TM 1051 1058 17 177
INEFIL NiMoCr  750  820  19  60
Thyssen UNION X85  790  880  16  53
Thyssen UNION X90  890  950  15  58
ESAB OK Tubrod 14.03 757 842 23 71
Thyssen UNION X96  930  980  14  40

Fig. 1 Graville diagram [20]

equivalent (CE). It can be clearly seen, that the examined Q + T base materials are
in the hardest weldable category (III.), which means that these steels require pre-
heating and we must control the linear energy precisely; while the TM base
materials locates in the easily weldable category (I.), so they do not require these
sort of measurements.
In the case of these steels, one of the most important features of the success-
fulness of the welding is the heat input, which can be described with the linear
energy. If the value of the linear energy is too low, the cooling rate of the welded
joint may be too fast, and then cold cracks can occur. In the opposite case, strong
coarse grain microstructure can be formed in the heat affected zone, which can
cause the decreasing of the strength and the toughness. Therefore, we received a
narrow welding lobe; inside this the quality of the joint may be suitable. Besides the
previously mentioned characteristics, the material quality, the chemical composition
and the applied thickness are important, too.
458 Á. Dobosy et al.

The t8.5/5 cooling time was used for the common description of the welding
conditions and parameters. Because of the previously mentioned reasons, the
cooling time is a narrow range in the case of the Q + T steels, but it can be changed
in a wide range in the case of the TM steels.

3.2 Welding Experiments

The welding parameters and the welding conditions were determined based on the
previously mentioned statements. For shielding gas, M21 designation with 18%
CO2 + 82% Ar composition was used, based on industrial experience. In the case
of the filler material 1.2 mm diameter wire was applied in all cases. The cases of the
base material-filler material pairing can be seen in Table 3. Further details of the
matching problems can be read in the literature [21–23]. In case of matching
condition, the evolved mechanical properties of the joints are equal or nearly the
same as the base material. In the undermatching case the mechanical properties of
the joints are lower, while in case of overmatching condition are higher than the
base material features.
The welding equipment was a DAIHEN VARSTROJ WELBEE P500L power
source. For the equal stress distribution X joint shape was used, with 80° opening
angle and with 1.5 mm gap between the two plates. During the welding, the test
pieces were rotated after each layer. The dimensions of the welded plates were
300 mm  125 mm. The root layers were made by a qualified welder; while the
other layers were made by automated welding car. The experimental assembly can
be seen in Fig. 2.
The applied welding parameters were summarized in Table 4. The table shows
the welding current (I), the voltage (U) and the welding speed (vh) values, also the
preheating (Te) and the interpass (Tr) temperatures, with the linear energy (Ev) and
the calculated critical cooling time (t8.5/5) values. The parameters of the root and the
filler layers were shown separately in each case.
In the case of S960TM material, wide ranges can be found due to the fact, that
we made experiments with small and large linear energy. It is important to note, that
both the interpass temperature and the linear energy are determining parameters in

Table 3 The base material-filler material pairing during our experiments

Base material Filler material Matching condition
S690QL INEFIL NiMoCr Matching (m)
Weldox 700E Thyssen UNION X85 Matching (m)
Weldox 700E Thyssen UNION X90 Overmatching (om)
S960QL Thyssen UNION X96 Matching (m)
S960QL ESAB OK Tubrod 14.03 Undermatching (um)
S960TM Thyssen UNION X96 Matching (m)
High Cycle Fatigue Investigations on High Strength Steels … 459

Fig. 2 Assembly of the welding experiments

Table 4 The applied welding parameters

Base material Layer Te/Tr, °C I, A U, V vh, cm/min Ev, J/mm t8.5/5, s
S690QL 1–2 150 130–140 19.0–20.5 20 700–750 7–8
3–8 180 280–300 27.5–28.5 40 1000– 9–11
1100
S960QL 1–2 180 120–130 18.5–19.0 20 570–590 5–6
3–8 150 280–305 28.5–31.0 45 900–1000 7.5–8.5
S960TM 1–2 60 130–140 19.0–20.5 20 700–750 7–8
3–8 150 260–310 25.0–32.0 30–60 900–1400 7–15

the case of the critical cooling time, however, between those two values there are no
close correlation.

4 High Cycle Fatigue Experiments and Their Results

4.1 Test Specimens

For the high cycle fatigue tests, flat specimens were used. The geometry of the
specimens and the cut out orientation from the base materials and the welded joints
460 Á. Dobosy et al.

Fig. 3 Specimen geometry

for the high cycle fatigue
experiments

Fig. 4 The orientation of the test specimens and the welded joints for the experiments

can be seen in Figs. 3 and 4, respectively. (Figure 4 is applicable not only for this
study, but also for our whole research work in this field.) In Fig. 3 the RD means
the rolling direction of the sheets, h is the longitudinal and k is the transversal
direction, furthermore, 1 and 2 mean the longitudinal and the transversal direction
of the welded joints, respectively. The thickness of the sheets (v) and the welded
joints (3) are perpendicular to the surface. Table 5 shows the nominal geometry of
the test specimens, where the orientations can be identified with Fig. 4.

4.2 Testing Circumstances

High cycle fatigue experiments were performed according to [24], using MTS
electro-hydraulic materials testing equipment (MTS 312 and MTS 810 (Fig. 5)), at
room temperature and on laboratory environment. Constant load amplitude was
applied during the experiments, with R = 0.1 stress ratio, f = 30 Hz loading fre-
quency, and sinusoidal loading wave form. Test specimens cut from the welded
joints were tested in as-welded condition.
High Cycle Fatigue Investigations on High Strength Steels … 461

Table 5 Main characteristics of the high cycle fatigue test specimens

Base Specimen axle/crack grow t, b, B, L, R,
material direction mm mm mm mm mm
S690QL BM h/v 6 8 24 175 92.5
GMAWm k/3 W 6 8 18 135 60
GMAWm k/1 W 6 7.5 18 135 60
Weldox BM h/k 6 7; 8 18 135 60
700E GMAWm k/1 W 6 7; 8 18 135 60
GMAWom k/1 W 6 8 18 150 60
S960QL BM h/k 6 8; 5.5 18 150 60
GMAWm k/3 W 6 7.75 18 150 60
GMAWum k/1 W 6 7.5 18 150 60
S960TM BM h/k 6 8 18 150 60
GMAWm k/1 W 6 8 18 150 60

Fig. 5 High cycle fatigue tests: a MTS 810 universal testing equipment; b the test specimen and
the grip

4.3 The Results of the High Cycle Fatigue Experiments

The results of the high cycle fatigue tests were shown in diagrams (Wöhler or S-N
curves).
The results and the S-N curves of the 690 strength category base materials
(amended with literature data [25–27]) can be seen in Fig. 6.
The results and the S-N curves of the 690 strength category gas metal arc welded
joints (with literature data [26, 27]) were summarized in Fig. 7.
462 Á. Dobosy et al.

1000
Stress range, Δσ, MPa

S690QL BM h/v [this study]

S690 BM R=0; f=3,5Hz [literature]
S690QL BM R=0; f n/a [literature]
Weldox700E BM h/k [this study]
Weldox700 BM R and f n/a [literature]
100
1,0E+03 1,0E+04 1,0E+05 1,0E+06 1,0E+07 1,0E+08
Number of cycles to failure, N t, cycle

Fig. 6 The results of the 690 strength category base materials

1000
Stress range, Δσ, MPa

S690QL GMAWm k/3W [this study]

S690QL GMAWm k/1W [this study]
S355-S690QL-S960QL GMAW butt weld; R=0; f n/a [literature]
Weldox700E GMAWm k/1W [this study]
Weldox700E GMAWom k/1W [this study]
Weldox700 butt weld; R and f n/a [literature]
100
1,0E+03 1,0E+04 1,0E+05 1,0E+06 1,0E+07 1,0E+08
Number of cycles to failure, Nt, cycle

Fig. 7 The results of the 690 strength category gas metal arc welded joints

The results and the S-N curves of the 960 strength category base materials
(integrated with literature data [27]) can be seen in Fig. 8.
Finally, the results and the S-N curves of the 960 strength category gas metal arc
welded joints (integrated with literature data [27]) were demonstrated in Fig. 9. In
High Cycle Fatigue Investigations on High Strength Steels … 463

Stress range, Δσ, MPa 1000

S960QL BM h/k [this study]

S960QL BM R=0; f n/a [literature]

S960TM BM h/k [this study]

100
1,0E+04 1,0E+05 1,0E+06 1,0E+07 1,0E+08
Number of cycles to failure, Nt, cycle

Fig. 8 The results of the 960 strength category base materials

1000
Stress range, Δσ, MPa

S960QL GMAWm k/3W [this study]

S960QL GMAWum k/1W [this study]
S355-S690QL-S960QL GMAW butt weld; R=0; f n/a [literature]
S960TM GMAWm k/1W t8/5=6-7s [this study]
S960TM GMAWm k1/W t8/5=16-17s [this study]
100
1,0E+04 1,0E+05 1,0E+06 1,0E+07 1,0E+08
Number of cycles to failure, Nt, cycles

Fig. 9 The results of the 960 strength category gas metal arc welded joints

the case of the S960TM steels, two kinds of experiments were made; in the first
case, the critical cooling time was t8.5/5 = 6–7 s, while in the second case it was
t8.5/5 = 16–17 s.
464 Á. Dobosy et al.

In the cases, where the test specimens did not fracture to a given cycle value
(usually 5 * 106 or 107 cycles), the results (marks in the diagrams) were denoted
with horizontal or a slope arrow.

5 Evaluation of the High Cycle Fatigue Experiments,

the Fatigue Design or Limit Curves

The performed high cycle fatigue experiments were compared with literature data;
accordingly we can evaluate the results more accurate. It is important to note, that
during some literature examinations R = 0 stress ratio was used, while for our
experiments R = 0.1 stress ratio was applied. This fact does not influence on our
conclusions significantly. Furthermore, the used frequencies during the literature
examinations are presumably smaller, than through our experiments; but the data do
not indicate significant differences (for example [13]), so this fact also does not
influence on our conclusions significantly. Finally, in the literature [27], the given
S355, S690QL and S960QL data are too general, so these values can use only for
general comparison.
The S690QL base material resistance against high cycle fatigue in thickness
direction is more favorable, than the Weldox 700E base material resistance in
transverse direction. The examined S690QL and Weldox 700E base materials were
shown better results, than the same literature data.
The matching (m) joints from the S690QL steel in the 3 W orientation have
lower resistance against the high cycle fatigue, than in the 1 W orientation; fur-
thermore, this steel has lower resistance than the Weldox 700E steel, in the 1 W
orientation. Both steel category have higher resistance against the high cycle fatigue
in the 1 W orientation, than the overall results of the butt welded joints, in the case
of the three steel category (S355, S690QL and S960QL [27]), in the 1 W orien-
tation. This is fully consentaneous with the approach, because in the former case
test specimens were used, and in the latter case the whole welded joints were
examined. The high cycle fatigue resistance, in the case of test specimens from the
S690QL steel matching (m) joints in 3 W orientation, is located in the range of the
three steel category butt welded joints in the 1 W orientation. The resistance of the
overmatched (om) joints against the high cycle fatigue in the case of the Weldox
700E steel is slightly higher, than the matched (m) joints resistance; but this
statement requires further investigations.
The high cycle fatigue resistance of the examined S960QL steel is better, than
the literature data in the same category. The high cycle fatigue resistance of these
steels is higher in the case of undermatched (um) joints, than in the case of
matching (m) joints. The high cycle fatigue resistance measured on specimens from
matched (m) and undermatched (um) joints are considerable higher (and indepen-
dent from the orientation), than the butt welded joints in the 1 W orientation, in the
High Cycle Fatigue Investigations on High Strength Steels … 465

case of the three steel category. This is consentaneous with the approach because of
the previously explained reason, but the differences are notable.
The high cycle fatigue resistance of the S960TM base material is more
favourable, than the examined S960QL base material, furthermore, than the overall
literature data. The results of matched (m) joints welded with the longer critical
cooling time (t8.5/5) are more advantageous, than welded with the shorter critical
cooling time.
Statistical method [24] was used for the numerical evaluation of the high cycle
fatigue experiments and for the determination of the fatigue design or limit curves.
Using logarithmic form and rearranging the expression (1) we get the following
equation:

log ðaÞ 1
logðDrÞ ¼  log ðNt Þ: ð2Þ
m m

This equation is a straight line on a double logarithmic scale, in our cases the
life-time region of the design curve. The parameters of the equation and other
characteristics of the curves can be found in Table 6. The Nk value is the number of
cycles for the break point of the S-N curve, the DrD is the fatigue limit, and the
Dr1E07 is the stress value belonging to 1 * 107 cycles in the cases, when the
horizontal (endurance limit) part of the curves cannot be determined.
As we can see in the Table 6, in four cases we cannot be determined the
endurance limit part of the curves. In two cases (S690QL BM h/v and S690QL
GMAWm k/1 W) the reason of this is the high resistance against the high cycle
fatigue. In the third case (Weldox 700E GMAWm k/1 W), the available data were
not enough for the determination of the second part of the S-N curve, while in the
fourth case (GMAWm k/1 W(16–17)), for the determination of the whole S-N curve.
In these cases the Dr1E07 stress values also cannot be determined.

Table 6 The numerical results of the high cycle fatigue experiments

Base Specimen axle/crack m log(a) Nk DrD Dr1E07
material grows direction – – Cycle MPa MPa
S690QL BM h/v 51.28 151.11 – – 646
GMAWm k/3 W 4.83 17.48 9.850E05 239 –
GMAWm k/1 W 50.25 141.26 – – 470
Weldox BM h/k 12.45 39.65 1.677E06 483 –
700 E GMAWm k/1 W 9.96 32.47 – – –
GMAWom k/1 W 17.45 52.93 2.320E06 466 –
S960QL BM h/k 10.29 33.89 1.014E06 513 –
GMAWm k/3 W 16.72 50.74 8.535E06 417 –
GMAWum k/1 W 12.59 40.18 4.944E06 456 –
S960TM BM h/k 11.51 37.90 6.995E06 500
GMAWm k/1 W(6–7) 7.52 25.81 1.369E06 412 –
GMAWm k/1 W(16–17) – – – 442 –
466 Á. Dobosy et al.

In accordance with these, the curves described by the given data in the Table 6,
can be explained as fatigue design or limit curves.

6 Summary and Conclusions

Based on the accomplished investigations, their results, the elaboration of the

results and the relevant literary data, furthermore, the performed comparisons the
following conclusions can be drawn.
• The results of the executed investigations justified the necessity of the statistical
approaches, especially referring to the directions of the welded joints and the
determination of the number of the tested specimens.
• Applying the developed welding technologies eligible welded joints can be
produced, where the appropriate quality contains the eligible resistance against
high cycle fatigue, too.
• The resistance of the base materials against high cycle fatigue is more advan-
tageous than the resistance of the welded joints. The quantity of the differences
in some cases suggests the possibility and the necessity of the modification of
the welding technology and/or the welding technological parameters.
• The matching problem has influence on the resistance of high strength steels
against high cycle fatigue, depending on the strength category.
• The resistance of the investigated base materials and welded joints against high
cycle fatigue is accordance with the resistance of the comparative materials can
be found in the literature, moreover, in some cases is better at it.

References

1. Steel: a key partner in the European low-carbon economy of tomorrow. European steel
technology platform (ESTEP), Brussels, March 2009, pp 1–16
2. European steel technology platform—Vision 2030. Report of the group of personalities.
European Commission, Luxembourg, March 2004, pp 1–35. ISBN:92-894-5036-3
3. Miller WS et al (2000) Recent development in aluminium alloys for the automotive industry.
Mater Sci Eng A 280:27–49
4. AluReport—AMAG customer and market information. 03. 2012, www.amag.at
5. Ghassemieh E (2011) Materials in automotive application. In: Chiaberge M (ed) New trends
and developments in automotive industry InTech, pp 365–394. www.intechopen.com. (ISBN
978-953-307-999-8)
6. Plastics and polymer composites technology roadmap for automotive markets. American
Chemistry Council, March 2014, pp. 1–59. www.americanchemistry.com
7. Metals Handbook, Volume 19. Fatigue and Fracture, ASM International, 1996
8. Balogh A, Dobosy Á, Frigyik G, Gáspár M, Kuzsella L, Lukács J, Meilinger Á, Nagy Gy,
Pósalaky D, Prém L, Török I (2015) Hegeszthetőség és a hegesztett kötések tulajdonságai:
High Cycle Fatigue Investigations on High Strength Steels … 467

Kutatások járműipari acél és alumíniumötvözet anyagokon, (Szerk.) Balogh A., Lukács J.,
Török I., Miskolc (Hungary), p 324. (ISBN 978-963-358-081-3)
9. Dobosy Á, Nagy Gy (2015) Különböző folyáshatárú acélok és hegesztett kötéseinek
kisciklusú fárasztása. XXII. OGÉT Nemzetközi Gépészeti Találkozó, Nagyszeben, 2014. 04.
24-27, pp 98–101, 2014 (ISSN 2068-1267)
10. Meilinger Á, Török I (2016) Lineáris dörzshegesztéssel készült kötések jellemzői kisciklusú
fárasztó igénybevétel esetén, GÉP LXVII. évf. 1. szám, pp 63–71
11. Lukács J, Meilinger Á, Pósalaky D Fatigue curves for aluminium alloys and their welded
joints used in automotive industry. Mater Sci Forum (In Press)
12. Lukács J, Nagy Gy, Harmati I, Koritárné FR, Kuzsella LnéKZs Szemelvények a mérnöki
szerkezetek integritása témaköréből, (Szerk.) Lukács J, Miskolci Egyetem Miskolc
(Hungary), pp 334, 2012 (ISBN 978-963-358-000-4)
13. Zsáry Á (1965) Méretezés kifáradásra a gépészetben. Műszaki Könyvkiadó, Budapest
14. Koncsik Zs, Lukács J (2013) Design curves for high-cycle fatigue loaded structural elements.
Mater Sci Forum 729:135–144
15. MSZ-EN 1993-1-1:2009: EUROCODE 3: Acélszerkezetek tervezése. 1-1 rész: Általános és
az épületekre vonatkozó szabályok
16. Stephens, RI, Fatemi A, Stephens RR, Fuchs HO (2001) Metal fatigue in engineering. Wiley
(ISBN 0-471-51059-9)
17. BS 7910:2013 + A1:2015: Guide to methods for assessing the acceptability of flaws in
metallic structures the British Standards Institution 2015. Published by BSI Standards Limited
2015 (ISBN 978 0 580 89564 7)
18. Barsom JM, Rolfe ST (1999) Fracture and fatigue control in structures: applications of
fracture mechanics. ASTM Manual Series: MNL41. American Society for Testing and
Materials, West Consthohocken, PA. (ISBN 0-8031-2082-6)
19. Lee YL, Pan J, Hathaway R, Barkey M (2005) Fatigue testing and analysis. Theory and
Practice. Elsevier Butterworth-Heinemann. (ISBN-10 0-7506-7719-8)
20. Dobosy Á, Gáspár M (213) Welding of quenched and tempered high strength steels with
heavy plate thickness. In: Proceedings 27th microCAD, international scientific conference,
Miskolc (Hungary), Paper M7. (ISBN 978-963-358-018)
21. Gáspár M, Balogh A (2013) GMAW experiments for advanced (Q + T) high strength steels.
Prod Process Syst 6(1):9–24
22. Balogh A, Gáspár M (2014) A matching kérdéskör: hozaganyagválasztás a konvencionális és
korszerű nagyszilárdságú acélok hegesztéséhez. Hegesztéstechnika 25(3):75–80
23. Gáspár M, Balogh A (2014) Behaviour of mismatch welded joints when undermatching filler
metal is used. Prod Process Syst 7(1):63–76
24. Nakazawa H, Kodama S (1987) Statistical S-N testing method with 14 specimens: JSME
standard method for determination of S-N curves. In: Statistical research on fatigue and
fracture. Current Japanese materials research. In: Tanaka T, Nishijima S, Ichikawa M
(eds) Elsevier applied science and the society of materials science, Japan, vol. 2, pp 59–69
(ISBN 1-85166-092-5)
25. Pijpers RJM, Kolstein MH, Romeijn A, Bijlaard FSK (2007) Fatigue experiments on very
high strength steel base material and transverse butt welds. Adv Steel Constr 5(1):14–32
26. Stemne D, Narström T, Hrnjez B (2010) Welding handbook. A guide to better welding of
Hardox and Weldox, 1st edn. SSAB Oxelösund AB. (ISBN 978-91-978573-0-7)
27. Hamme U, Hauser J, Kern A, Schriever U (2000) Einsatz hochfester Baustähle im
Mobilkranbau. Stahlbau 69(4):295–305
Toughness Examination of Physically
Simulated S960QL HAZ by a Special
Drilled Specimen

Marcell Gáspár, András Balogh and János Lukács

Abstract Based on the welding heat cycle models physical simulators are capable
for the creation of critical heat-affected zones (HAZ). The simulated HAZ areas can
be examined by various material testing methods (e. g. Charpy V-notch impact test)
due to their increased homogeneous volume compared to their extension in real
welding experiments. In our research work relevant technological variants (t8.5/5 =
2.5…30 s) for gas metal arc welding technology were applied during the HAZ
simulation of S960QL steel (EN 10025-6), and the effect of cooling time on the
coarse-grained HAZ was analysed. In thermo-mechanical simulators the achievable
cooling rate is always the function of specimen geometry and the presence of
external cooling. Therefore a special drilled specimen with external cooling was
developed for performing a shorter (t8.5/5 = 2.5 s) cooling than 5 s, which cannot be
realized on the conventional Gleeble specimen. Heat cycles were determined
according to the Rykalin 3D model. The properties of the selected coarse-grained
(CGHAZ) zone were investigated by scanning electron microscope, hardness test
and Charpy V-notch impact test.

1 Introduction

Quenched and tempered (Q+T) high strength steels, applied in mobile structures
(e.g. trucks, mobile cranes) belong to the highest category of structural steels. Their
spreading application in vehicle industry is motivated by their outstanding strength

M. Gáspár (&)
A. Balogh
J. Lukács

University of Miskolc, Miskolc, Hungary
e-mail: gasparm@uni-miskolc.hu
A. Balogh
e-mail: metweld@uni-miskolc.hu
J. Lukács
e-mail: janos.lukacs@uni-miskolc.hu

© Springer International Publishing AG 2017 469

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_40
470 M. Gáspár et al.

and toughness characteristics. Due to their outstanding mechanical properties sig-

nificant weight reduction can be achieved by their application.
These steels have a non-equilibrium tempered martensitic (in lower strength
categories tempered martensitic + bainitic) microstructure due to the water cooling
applied in the quenching cycle and to the high temperature tempering [1]. In order
to realize the quenching condition in the whole cross section, alloying components
(Cr, Mo) are added to the steel, which move the CCT curves to the right.
Microalloying elements (Nb, V and Ti) are also used in order to ensure a fine grain
microstructure and to prevent grain coarsening. The examined S960QL [2] steel has
a guaranteed 27 J Charpy-V energy at −40 °C thanks to the fine grained
microstructure (approx. 10 µm grain size) and low impurity level (minimized S and
P content). The excellent strength of Q + T high strength steels is originated pri-
marily from the fine tempered lath-like martensitic microstructure (Fig. 1) and the
dislocation density. The effect of the dissolved carbon is generally lower [3].
The chemical composition and the calculated carbon equivalents of the inves-
tigated base material according to the material certification is summarized in
Table 1.
The required and the measured mechanical properties of the examined S960QL
heat according to the material certification is summarized in Table 2.
The strength and toughness characteristics of the base material can significantly
decrease during the welding heat input due to the irreversible microstructural

Fig. 1 Tempered martensitic microstructure of the base material, Nital 2%

Toughness Examination of Physically Simulated S960QL HAZ … 471

Table 1 Chemical composition and carbon equivalents of S960QL in mass percent

S960QL C Si Mn P S Cr Ni CEV
0.17 0.23 1.23 0.011 0.001 0.20 0.04 0.54
Mo V Ti Cu Al Nb B N CET
0.588 0.041 0.004 0.01 0.061 0.017 0.001 0.002 0.36

Table 2 Mechanical characteristics of the investigated S960QL

Rp0.2 Rm Rp0.2/Rm A5 KV (at −40 °C)
MPa MPa – % J
 960 980–1150 –  10  27
1014 1053 0.96 14 75

changes in the heat-affected zone. Besides softening hardness peaks can also locally
form in the HAZ. Next to the fusion line the material is heated much above Ac3
temperature therefore homogeneous austenite forms. When a coarse-grained zone
(CGHAZ) forms the peak temperature is above 1100 °C where the grains start to
exponentially expand in the function of the presence of different microalloying
elements [4]. The decreased toughness of this zone has two reasons in quenched
and tempered high strength steels. On the one hand the grain size can approximately
be more than 10 times higher than the base material (>100 µm). Another reason
derived from the alloying elements, resulting martensite with rough lath-like
structure at shorter cooling times. Long cooling times may result coarse,
upper-bainitic microstructure where M-A constituents can occasionally form,
resulting in extremely low impact energy values [5]. Besides the weld metal
CGHAZ has the highest risk of cold cracking since the hydrogen can diffuse from
the fusion line to the brittle, coarse-grained microstructure resulting cold cracks
with the residual tensile stress of the welded joint. Microalloying elements (in the
present case Nb and V) can form small, disperse precipitates at the grain boundaries
restraining excessive grain growth [4].
In real welding experiments the critical parts (e.g. CGHAZ) of the heat-affected
zone is limitedly identified and investigated due to their narrow extension.
Nowadays, physical simulation opened a wide range of examination possibilities
for the precise analysis of heat-affected zone [6]. In the presented experimental
work HAZ tests have been performed in a new generation of simulators, called
Gleeble 3500, installed at the Institute of Materials Science and Technology of the
University of Miskolc (Fig. 2). By the application of HAZ test the desired part of
the heat-affected zone can be precisely and homogeneously created in a volume
sufficient for the further material tests, e.g. Charpy V-notch impact test.
472 M. Gáspár et al.

Fig. 2 Gleeble 3500 thermo-mechanical, physical simulator installed at the University of Miskolc

2 Experimental Plan

In our research work the aim was to analyse the effect of arc welding parameters on
the HAZ properties of S960QL. Although more welding heat cycle models are
available in the QuickSim software developed for the simulator, in the present case
the GSL programs were manually written, using the time and temperature points of
the total heating and cooling phase calculated by the Rykalin 3D model [7]. This
model describes the temperature field generated by a moving point-like heat source
on the surface of a semi infinity body. In this case 3D thermal conductivity is
dominant while surface heat transfer (convection) is negligible. There were more
reasons for the selection of 3D model. Firstly, this equation is independent from
plate thickness, therefore the number of variables can be decreased during the
analysis. Secondly the investigated S960QL is often used in medium and heavy
plate thickness (>15 mm) where the 3D model may give more precise result.
Thermophysical properties were determined for the whole relevant temperature
range (between 20 and 1400 °C) by the application of JMatPro software, and
then their average values were used in the Rykalin equation (k = 37.8 W/(m°C),
cp = 690.2 J/(kg°C), q = 7614.7 kg/m3). JMatPro software was applied at the
University of Oulu during the internship of one of the authors. Preheating and
interpass temperatures (T0) were set to 150 °C according to our previous welding
experiments [8].
Our aim was to simulate the CGHAZ of the investigated S960QL for a t8.5/5
cooling range (cooling time from 850 to 500 °C), typical for the generally applied
gas metal arc welding (GMAW) process: 2.5…30 s. Short cooling times (2.5…5 s)
can be relevant at root welding or high speed robotic welding. Longer cooling times
Toughness Examination of Physically Simulated S960QL HAZ … 473

Fig. 3 CGHAZ heat cycles 1400

between t8.5/5 = 2.5 and 30 s
1200
2.5 s
1000 5s
850 ºC 10 s
800 15 s

T, ºC
22.5 s
600 30 s
500 ºC

400

200

0
0 50 100 150 200 250
t, s

(>20 s) may also happen in the industrial practice when weaving technique is
applied by the welder. Considering the selection of peak temperatures our moti-
vation was to generate the most critical part of CGHAZ, having the lowest
toughness. Hence, the peak temperature was selected for 1350 °C in order to
produce the occurring largest grains. This value is safely lower than the nil-strength
temperature of the investigated steel (NST = 1408 °C [9]).
The designed CGHAZ cycles for the technological variants between 2.5 and 30 s
are illustrated in Fig. 3.
Before running the simulation in Gleeble the optimal specimen geometry should
be also selected. The cooling rate is generally determined by the copper grips and
the water-cooled jaws surrounding the specimen. The specimen geometry and the
application of external cooling can also significantly affect the cooling. At HAZ test
when Charpy V-notch impact tests are planned, the conventional experimental
set-up demands 10  10  70 mm square bar specimen. By the application of this
specimen geometry the achievable shortest cooling time is approximately
t8. 5/5  5 s (around 70 °C), whilst in the case of the GMAW shorter cooling time
may happen. Therefore, there was a demand for a new specimen geometry capable
of determining the impact energy of CGHAZ at short cooling times.

3 Development of a Special Drilled Specimen

3.1 Preliminary Tests

Since there is a technical possibility in Gleeble 3500 for the application of external
cooling, the development of a drilled specimen geometry was planned which can be
connected to the cooling systems through the holes on its both ends. It is important
to note that cylindrical drilled specimen with a reduced cross section at the middle
474 M. Gáspár et al.

is generally applied for the determination of CCT diagrams where the extremely
high cooling rate is often needed. The cooling set of this cylindrical specimen is
planned to be applied to the developed new specimen geometry.
The first step was to analyse the effect of the holes at the end of the standard
Charpy V-notch 10  10  55 mm specimen, which will be manufactured from
the planned new Gleeble specimen. It is obvious that the diameter and the depth of
the holes may affect the fracture during the impact test. Therefore, standard Charpy
specimens were manufactured from the investigated base material, and holes were
drilled with several depths (l) and diameter (d) sizes. Three specimens were pre-
pared from each geometry. Then the tests were performed at −15 °C by a PSD
300/150 type impact testing equipment. The results of the absorbed energy (CVE)
measured on the modified Charpy specimen (Fig. 5) are summarized in Table 3.
The fracture surface of the specimens was also analysed. In case of the deepest
hole (l = 25 mm) the crack initiated at the radius of the V-notch, but then propa-
gated partially towards the holes. When l = 22.5 mm was applied the hole did not
have an effect to the crack propagation and hence to the impact energy. By con-
sidering that notable difference was not observed between the tested 6 and 7 mm
hole diameter, the latter physical simulation specimens have d = 7 mm in order to
realize a more intensive cooling.

3.2 Geometry

After performing the preliminary tests the specimen geometry was finalized
(Fig. 4), and the necessary specimens were manufactured from the S960QL plate.

Table 3 The effect of the hole geometry on CVE at −15 °C

d, mm l, mm CVE1, J CVE2, J CVE3, J CVEmean, J D, J
6 0 133 160 133 142 15.6
17.5 145 81 172 133 46.7
20 176 187 201 188 12.5
22.5 128 173 165 155 24.0
25 146 117 94 119 26.1
7 22.5 136 148 149 144 7.2

Fig. 4 The developed Gleeble specimen for the analysis of HAZ toughness at short cooling times
Toughness Examination of Physically Simulated S960QL HAZ … 475

After the successful Gleeble tests the specimen geometry has to be modified
(Fig. 5) according to the requirements of Charpy V-notch impact test.

3.3 Test Results

Generally, one thermocouple is enough for the process control, but in present case,
two K-type (NiCr–Ni) thermocouples were welded to the opposite sides of the
specimens at the middle in order to control whether the heat input symmetrical
enough.
Then the specimens were placed into the grips and the cooling device was joined
at the two sides (Fig. 6) before fixing it into the jaws.
Free span between the grips was set to 7.5 mm in order to maximize the cooling
rate without wagering the success of the simulation (Gleeble Application Note
allows even 5 mm, although the optimal is around 10 mm). Compressed air was
applied to the cooling, although water can also be used.

Fig. 5 Modified Charpy V-notch specimen

Fig. 6 Drilled square bar

specimen joined to the
cooling device in the copper
grips
476 M. Gáspár et al.

Fig. 7 Programmed (PTemp) 1400

and measured (TC1 és TC2) t8,5/5 = 2,5 s PTemp
heat cycles 1200
TC1
1000
850 ºC TC2
800

T, ºC
600 500 ºC
400

200

0
0 10 20 30 40
t, s

After the first simulations inhomogeneity was noticed in the grain size during the
microscopic tests. The grains closer to the surface had larger size compared to
the inner part, so the maximum temperature was not equal at the middle and the
surface. After further experiments the conclusion was to apply a 2 s holding time at
the peak temperature, which can equalize the temperature between the middle and
the surface. Therefore the heat cycle of the t8.5/5 = 2.5 s (Fig. 3) had to be modified,
and only the cooling part was described by the Rykalin 3D. Heating rate was set to
1000 °C/s, followed by a 2 s holding time. The programmed and the measured
thermal cycles by the two thermocouples are illustrated in Fig. 7.
In Fig. 7 the red (TC1) and blue (TC2) curves, indicating the measured heat
cycles, nearly overlapped the programmed cycle (black curve). This verifies that by
the application of this special drilled specimen with external compressed air cooling
t8.5/5 = 2.5 s can be achieved, which is a double cooling rate compared to the
conventional specimen set-up at HAZ test.

4 Material Tests

4.1 Microscopic Tests

After the physical successful simulations with the two specimen geometries
(10  10  70 mm for 5…30 s, and special geometry presented in Fig. 4 for
2.5 s) the samples were perpendicularly cut at the thermocouples and the surfaces
were prepared for the microstructural analysis. The properties of CGHAZ and the
effect of welding parameters were examined with a ZEISS EVO MA10 scanning
electron microscope. Samples were coated with a thin gold layer in order to develop
picture quality. The microstructures of CGHAZ at 2.5, 5 and 30 s are presented
between Figs. 8, 9 and 10.
All microscopic tests verified that the desired CGHAZ was successfully simu-
lated in all cases. Regarding CGHAZ rough lath-like martensitic microstructure
Toughness Examination of Physically Simulated S960QL HAZ … 477

Fig. 8 CGHAZ microstructure at t8.5/5 = 2.5 s, Nital 2%

Fig. 9 CGHAZ microstructure at t8.5/5 = 5 s, Nital 2%

478 M. Gáspár et al.

Fig. 10 CGHAZ microstructure at t8.5/5 = 30 s, Nital 2%

with large (>100 µm) prior austenite grain size was noticed between the investi-
gated whole cooling time range: t8.5/5 = 2.5…30 s. Occasionally the measured grain
size was even 200 µm. The grain size increased with the longer cooling times. It is
important to note that in physical simulations the grain size of CGHAZ is generally
larger than in the real welded joints [6]. The type of microstructure did not change
significantly in the function of the cooling time, although the characteristics of the
self-tempering of martensite were noticed marked by small carbide precipitates at
t8.5/5 = 30 s, where the conditions of carbon diffusion are more favourable. The
results correlated with the JMatPro calculations.

4.2 Hardness Tests

A Reichter UH250 universal macro hardness tester was applied for the examination.
Five measurements were done on the cut surface. The average macro hardness
values of the simulated CGHAZ are summarized in Fig. 11.
An evaluation was performed according to the EN 15614-1 [10] standard, which
permits HVmax = 450 HV10 for the non-heat treated welded joints (including
HAZ) of quenched and tempered high strength steels belonging to the 3rd group of
CR ISO 15608. In Fig. 11 340 HV10 indicates the average hardness of the base
material. In the investigated cooling time range CGHAZ fulfilled the requirement of
the governing standard, and critical hardening or softening there was not noticed.
Toughness Examination of Physically Simulated S960QL HAZ … 479

500

450
y = 459,7x-0,048
R² = 0,9624
400
HV10

350
340

300

250

200
0 5 10 15 20 25 30 35
t 8.5/5, s

Fig. 11 Hardness of CGHAZ in the function of t8.5/5 cooling time, HV10

The maximum hardness (437 HV10) was measured at t8.5/5 = 2.5 s in CGHAZ. The
effect of cooling time can be clearly identified with the hardness values of the
CGHAZ specimens. The average hardness reduced as the cooling time was longer.
Although the requirement was fulfilled in CGHAZ it can be important to mention
that in quenched and tempered high strength steel the fine-grained zone (FGHAZ)
can have similarly high hardness, which may exceed the CGHAZ hardness [11].

4.3 Charpy V-Notch Impact Tests

Minimum five specimens from each heat cycle were used for the impact tests per-
formed in PSD 300/150 equipment. According to EN 10025-6 the required mini-
mum impact energy is 27 J at −40 °C for S960QL. According to the material
certificate the investigated steel plate had 75 J average impact energy, although 167 J
was measured during our impact test on the base material. The individual and the
mean CVE values are illustrated in Fig. 12.
It can be concluded that all of the investigated heat-affected zones in the
examined t8.5/5 cooling time interval had significantly lower toughness than the base
material. In almost all cases except t8.5/5 = 30 s there were individual values that
did not reach the required minimum level. Above t8.5/5 = 15 s the impact energy
slightly increased by the cooling time when martensitic microstructure was noticed
that may be originated from the self-tempering effect presented at the microstruc-
tural analysis. Although the average absorbed energy was slightly above the
required 27 J at almost every cooling time (except t8.5/5 = 15 s), the specimens
fractured brittle, which means that 27 J requirement cannot fully guarantee the
ductile behaviour of the heat-affected zone.
480 M. Gáspár et al.

70
1

60 2
3
50 4
CVE, J [-40 0C]

5
40
6

30 7

27 8
20 9
10
10
11

0 12
0 5 10 15 20 25 30 35 Mean
t8.5/5, s

Fig. 12 Charpy-V energy of CGHAZ in the function of the cooling time

5 Conclusions

The CGHAZ of S960QL was successfully performed by the Gleeble 3500 physical
simulator between t8.5/5 = 2.5…30 s cooling time.
For the physical simulation and the toughness analysis of t8.5/5 = 2.5 s a drilled
specimen geometry was developed and tested.
Comparing to mild steels (e.g. S355J2+N) quenched and tempered high strength
steels are more sensitive to the welding heat input. In the CGHAZ of S960QL
relatively high (>400 HV10) hardness values were measured between t8.5/5 = 2.5…
22.5 s. However, the measured macro hardness values did not exceed the 450 HV10
requirement of the governing standard.
In the examined t8.5/5 cooling time interval CGHAZ had significantly lower
toughness (the mean was slightly above 27 J at −40 °C) than the base material.

Acknowledgements The research work presented in this paper based on the results achieved within
the TÁMOP-4.2.1.B-10/2/KONV-2010-0001 and TÁMOP-4.2.2.A-11/1/KONV-2012-0029 pro-
jects in the framework of the New Széchenyi Plan. The realization of these projects was supported by
the European Union, and co-financed by the European Social Fund.

References

1. Weglowsky M (2012/2) Modern toughened steels—their properties and advantages, Biuletyn

Instytutu Spawalnictwa, pp 25–36
2. EN 10025-6: Hot rolled products of structural steels. Technical delivery conditions for flat
products of high yield strength structural steels in the quenched and tempered condition
Toughness Examination of Physically Simulated S960QL HAZ … 481

3. Porter DA, Easterling KE (1996) Phase transformations in metals and alloys, 2nd edn.
Chapman and Hall, 2-6 Boundary Row, London SE1 8HN, UK. (ISBN 0-412-45030-5)
4. Bhadesia HKDH, Honeycombe RWK (2006) Steels microstructure and properties, 3rd edn.
Elsevier Linacre House, Jordan Hill, Oxford OX2 8DP, UK
5. Nevasmaa P (1996) Evaluation of HAZ toughness properties in modern low carbon low
impurity 420, 550 and 700 MPa yield strength thermomechanically processed steels with
emphasis on local brittle zones. Lisensiaatintyö, University of Oulu, p 176
6. Adony Y (2006) Heat-affected zone characterization by physical simulations. Welding J 42–47
7. Rykalin NN (1953) Teplovüe processzi pri szvarke, Vüpuszk 2 Izdatelsztvo Akademii Nauk
SzSzSzR, Moszkva, p 56
8. Gáspár M, Balogh A (2013) GMAW experiments for advanced (Q+T) high strength steels.
Prod Process Syst 6(1). University of Miskolc, Department of materials processing
technologies, pp 9–24
9. Kuzsella L, Lukács J, Szűcs K (2013) Nil-strength temperature and hot tensile tests on
S960QL high-strength low-alloy steel. Prod Process Syst 6(1):67–78
10. EN ISO 15614-1: Specification and qualification of welding procedures for metallic materials—
welding procedure test—Part 1: arc and gas welding of steels
11. Dunne DP, Pang W (2015) Heat affected zone hardness of welded low carbon, quenched and
tempered steels. IIW IX-2523-15
Innovation Methods for Residual Stress
Determination for the Automotive
Industry

Máté Sepsi, Dávid Cseh, Ádám Filep, Márton Benke

and Valéria Mertinger

Abstract Determination of the residual stress state in a loaded automotive com-

ponent is highly important because of its strong effect on the lifetime of the element.
Nowadays, the residual stress characterization of products became an everyday
requirement in the automotive industry, and the quality control is impossible to
imagine without it. Forasmuch as every producing process (casting, heat treating,
different kinds of metal deformation processes and surface compressing methods,
etc.) influences the residual stress state, therefore it can be very complex and
various within the materials. If we conscious in the effect of these processes it is
possible to reach such a state in the material which can enhance its lifetime and
quality, and with an optimized process, the costs could be reduced. Several methods
exist for measuring the residual stress and each of them has its own advantage and
disadvantage. With this paper our purpose is to introduce nowadays’ available
stress measuring methods such as X-ray diffraction, magnetic Barkhausen noise and
the hole drilling methods and a few more alternatives. Some useful results from the
practice are also presented.

M. Sepsi (&)
D. Cseh
Á. Filep
M. Benke
V. Mertinger

Institute of Physical Metallurgy, Metalforming and Nanotechnology,
University of Miskolc, Miskolc, Hungary
e-mail: mate.sepsi@gmail.com
D. Cseh
e-mail: cseh88david@gmail.com
Á. Filep
e-mail: adam9032@gmail.com
M. Benke
e-mail: fembenke@uni-miskolc.hu
V. Mertinger
e-mail: femvali@uni-miskolc.hu

© Springer International Publishing AG 2017 483

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4_41
484 M. Sepsi et al.

1 Introduction

Residual stress is introduced into solid state materials during almost all shape
modifying processing types including casting, plastic deformation, machining,
welding, etc. For most cases, the presence of residual stress is not designed, but is a
consequence of the required manufacturing step [1, 2]. In the automotive industry,
however, residual stress is introduced within a machine component intentionally to
enhance the component’s properties, such as resistance against fatigue [3, 4]. The
most common techniques to induce compressive stresses are shot peening and ball
burnishing. In such cases, the stress state (type and magnitude) of the component is
a requirement and regulated similarly to mechanical properties or geometry.
Because of that, providing accurate data about residual stress is of great interest.
There are many methods that can qualitatively describe the residual stress state of a
component [5, 6]. One of them is the Magnetic Barkhausen Noise measurement,
which could be the quickest and the easiest measurement from the side of the
operator. It is based on a ferromagnetic phenomenon which induced by a changing
magnetic field in a ferromagnetic material. Forasmuch the characteristic of the
Barkhausen noise is influenced by the microstructure, the hardness and the stress
state of the magnetized material, therefor it is suitable to use it in the industry for
quality control, where the measurement time is very important [7]. Nevertheless,
the description of the residual stress state quantitatively is not easy with this method
[1]. Another one is the Hole-drilling method, where the definition of the residual
stress based on the macroscopic deformation of a solid material upon changing the
stress balance through material removal. The largest advantage of this kind of
measurement is that not just the ferromagnetic and the polycrystalline materials are
analysable with it. But, as it is mentioned, material should be removed to get result,
so stress state on the surface is not definable. However, diffraction based methods
are the only ones that provide accurate quantitative information about the stress
state from the surface to the deep. While neutron diffraction requires a neutron
source, X-ray diffraction based residual stress measurement can be performed in
laboratory conditions, or, even as on-site measurements. Using non-destructive
X-ray diffraction, no sample has to be cut from the component for the examination.
Using this technique, not only remains the component usable after the examination,
but the true stress state is not distorted with sample cutting. Thus, the true stress
state is provided after such examination. The residual stress measurement by X-ray,
as all of the other diffraction measurements is based on Bragg’s law [8]:

nk ¼ 2d sin h ð1Þ

This equation gives the diffraction angle (h) where the diffracted X-ray photons
with a given wavelength (k) meet in the same phase after scattered on the lattice
with a given plane distance (d), (n is an integer.) An intensity peak appears as a
function of the diffraction angle in the recorded interference function where Bragg’s
law is fulfilled. The diffraction angle of a stressed material is shifted relative to the
Innovation Methods for Residual Stress Determination … 485

relaxed value. The lattice parameter change is the source of this phenomenon which
is generated by the residual stress. Assuming linear elastic distortions, the related
normal stresses can be calculated by Eq. (2) which is called sin2w method [8]:

dw  d0 E
r¼  ð2Þ
d0 ð1 þ mÞ sin2 w

where d0 and dw are the lattice plane distances of the strained (stressed) material in
the normal and in the tilted position defined by the angle w, respectively. E (Young
modulus) and m (Poisson number) are the elastic parameters of the material. In
practice, several tilting positions (± () in one measuring point are usually applied
and after the interferential functions are recorded, d), normal stress and its scatter,
shear stress and its scatter, the average 2 (values and the average full with half
maximum (FWHM) can be calculated. Figure 1 provides the process of an XRD
residual stress data in one measuring point. In this case 5 tilt positions (3/3) were
applied and the interferential functions were recorded at each (W = −45, −30, 0,
+30, +45°) tilt positions. The sloop of the parabola axis on the plot of calculating d
as the function of sin2W gives the value of normal stress data. The direction of the
measured stress is given by the direction of the applied tilting was used during the
measurement. The FWHM and the Bragg angle are calculated as an average value

Fig. 1 The process of an XRD residual stress data. The recorded interferential functions at
different (W = −45, −30, 0, +30, +45°) tilt positions. The plot of calculating d as the function of
sin2W. The gears under the measuring process at the A location. The calculated results are: r:
−448.7 ± 9.6 MPa, s: −16.0 ± 2.1 MPa, FWHM: 4.57° ± 0.07°, 2h:155.77°
486 M. Sepsi et al.

of the data of interferential functions which were recorded at different tilting

positions.
The X-ray diffraction technique is a reliable method for stress measurements.
However, the residual stress state of a component can be measured accurately only
if the true stress state is not affected by sample cuts. This requirement demands a
special equipment and measurement method [2, 8]. Non-destructive X-ray
diffraction is generally accepted as the ultimate tool for residual stress measure-
ments and the method meets the requirements of the regarding Machinery and Low
Voltage directives and applies the following standards: SFS-EN ISO 12100-1,
12100-2, 13857, 60204-1. Even the method is standardized, the effect of mea-
surement parameters on the obtained result is still debated. The objective of the
present manuscript is to examine the effect of measurement parameters, namely
tilting range, number of tilting positions and spot size of the measured stress values
of a shot-peened gear wheel. Our results also give evidence about the repro-
ducibility and sensitivity of the diffraction method.

2 Experimental

The residual stress was measured non-destructively by a Stresstech Xstress 3000

G3R type centerless X-ray diffractometer, owned by the Institute of Physical
Metallurgy, Metalforming and Nanotechnology at the University of Miskolc. This
method provides residual stress values from *10 lm depth of the surface layer
(depending on material type, applied generator power, Bragg-angle). A Cr radiation
was used to measure the {211} reflections of the samples. For stress calculations,
the following material constants were used: Poisson’s ratio m = 0.3, Young’s
modulus E = 211000 MPa. A carburized and shoot peened gear wheel was chosen
as test component. All of the measurements were performed at the same two
locations (D and A location) and the same direction of the gear wheel. The mea-
sured parameters (the investigated variables) such as maximum tilting angle, the
number of tilting and spot size (the diameter of the beam on the examined surface)
were systematically changed during the test. The exposure time was chosen based
on the requirement of proper signal/background ratio, so it was changing, but it is
not the investigated variable in this paper. The spot size was 1, 3, 5, mm in
diameter. The maximum tilting angles were ±20° or ±40°. The 3/3, 5/5, 7/7 and
11/11 number of tilts were chosen which refers that one stress data were calculated
based on 3, or 5, or 7, or 11 tiltings in both plus and minus directions. Three parallel
measurements were performed in each case of the measuring conditions. Thus,
reproducibility was also tested.
Innovation Methods for Residual Stress Determination … 487

3 Results and Discussion

Figure 1 provides one of the measured data at location A as an example and

Table 1 summarizes all of the calculated data, such as normal stress (r), shear stress
(s), FWHM and Bragg angle (2h). The normal and shear stresses refer to the stress
state in the measured point and direction, which was perpendicular to the axis.
The FWHM value is larger if the micro stress caused by dislocations or interstitials
is also larger. The Bragg angle is associated with the difference of chemical
composition, especially the carbon solution of the carburized gear surface zone.
Figures 2, 3, 4, 5, 6, 7, 8, 9 and 10 show the results of Location A and D which
were the equivalent parts of the gear. The error bars of each data are also presented.
The significant different data of the two locations are originated from the treatment
of the gear which can also be important information. The sufficient number of tilts
has been always a question for the users. Too high number increases the measuring
time while the too low causes wide error bars. Figures 2 and 3 show that it is a false
theory. The stress results of 11 tilts number shows only smaller error bar, but the
difference is much smaller than the difference of the result from the equivalent parts
(location A and D) of the gear. The FWHM data show no difference.
The optimal spot size is also a regular question for the operators. The larger
collimator (larger spot size) gives higher reflected intensity so the measuring time is
decreased which is convenient situation. Also, a feeling of the users that using
larger collimator gives better statistic for the material results that the commonly
used spot size 3–4 mm in diameter. As we see from Figs. 4 and 5 the stress data
deviation is within the error bars, even if 6 times larger spot size was used, so the
thumb rule works, but the FWHM data are strongly increased as the spot size is
increasing. Which does not originate from the microstructure, but is an artificial
effect. So the FWHM data can be compared in the case of same spot size mea-
surements only.
The result in Figs. 6, 7, 8 and 9 has a serious message. The ±45° maximum
tilting angle is usually used, but in some cases the special geometry limit does not
yield such a big tilting. Measuring in the bottom land of a gear is a typical example.
In that case the operators cut the gears or apply different measuring mode (psi or
omega) or decrease the maximum tilting angle. As we have already mentioned in
the introduction, the sample cutting can modify the stress state so the measured data
do not represent the real initial situation. The question is always the effect which
generated the stress state and the volume fraction of the sample where this state is in
equilibrium. Applying the omega mode can be a good solution, but some of the
available equipment are not able to fulfil this. Figure 6 shows that applying ±20°
tilting angle instead ±45° results an extremely large increase in the error bars of the
measured points and also the large scatter of the parallel measurements and the spot
size variation does not help at all (Fig. 7). In front of this, it is obvious that the
FWHM data unaffected by the tilting angles (Fig. 8) only the spot size effect is
detectable (Fig. 9).
Table 1 Calculated data
488

Location Var. No. Tilt max, ″ No. of tilt Spot size, mm r, Mpa (±) s, MPa (±) FWHM, ° (±) 2h, °
Location D a 1 45 3 3 −444.20 11.60 −10.30 2.60 4.52 0.09 155.76
a 2 45 3 3 −448.70 9.60 −16.00 2.10 4.57 0.07 155.77
a 3 45 3 3 −435.90 14.50 −11.60 3.20 4.53 0.10 155.80
b 4 45 5 3 −437.90 7.90 −7.10 1.60 4.53 0.06 155.79
b 5 45 5 3 −443.90 4.80 −7.30 1.00 4.53 0.07 155.76
b 6 45 5 3 −447.40 8.40 −8.50 1.70 4.51 0.06 155.76
c 7 45 7 3 −443.60 5.90 −7.30 1.10 4.50 0.06 155.80
c 8 45 7 3 −432.50 5.20 −7.60 1.00 4.50 0.07 155.76
c 9 45 7 3 −452.10 7.50 −7.30 1.40 4.52 0.07 155.77
d 10 45 11 3 −450.10 6.00 −9.40 1.10 4.51 0.06 155.77
d 11 45 11 3 −448.50 4.70 −8.20 0.90 4.50 0.06 155.77
d 12 45 11 3 −441.70 5.70 −9.40 1.10 4.51 0.08 155.77
e 13 20 5 5 −459.50 30.50 −1.70 2.40 4.43 0.04 155.76
e 14 20 5 5 −445.60 28.80 −5.90 2.30 4.46 0.04 155.76
f 15 45 5 3 −437.90 7.90 −7.10 1.60 4.53 0.06 155.79
f 16 45 5 3 −443.90 4.80 −7.30 1.00 4.53 0.07 155.76
f 17 45 5 3 −447.40 8.40 −8.50 1.70 4.51 0.06 155.76
g 18 20 5 3 −459.50 30.50 −1.70 2.40 4.43 0.04 155.76
g 19 20 5 3 −445.60 28.80 −5.90 2.30 4.46 0.04 155.76
h 20 45 5 3 −438.4 11.30 −9.10 2.20 4.53 0.08 155.77
h 21 20 5 3 −438.3 35.80 −7.70 2.80 4.44 0.03 155.76
h 22 20 5 3 −445.3 27.70 −5.40 2.20 4.48 0.03 155.76
i 23 20 5 1 −429.4 26.90 5.30 2.10 4.36 0.03 155.83
i 24 20 5 1 −408 27.60 2.70 2.20 4.35 0.03 155.84
i 25 20 5 1 28.90 3.40 2.30 4.36 0.04 155.81
M. Sepsi et al.

−464.5
(continued)
Table 1 (continued)
Location Var. No. Tilt max, ″ No. of tilt Spot size, mm r, Mpa (±) s, MPa (±) FWHM, ° (±) 2h, °
Location A 1 20 5 5 −498.4 16.80 11.50 1.30 5.06 0.03 155.71
2 20 5 5 −448.6 11.10 −10.10 0.90 5.07 0.02 155.71
3 20 5 5 −504.7 15.60 −10.00 1.20 5.08 0.03 155.72
4 20 5 4 −481.9 28.30 −8.40 2.20 4.91 0.05 155.71
5 20 5 4 −446.9 22.60 −10.60 1.80 4.91 0.03 155.71
6 20 5 4 −503.1 16.20 −8.40 1.30 4.91 0.02 155.72
7 20 5 3 −467.6 16.00 −6.80 1.30 4.73 0.03 155.73
8 20 5 3 −492.6 28.20 −7.60 2.20 4.73 0.03 155.73
9 20 5 3 −463 20.80 −6.60 1.60 4.73 0.04 155.73
10 20 5 1 −485.2 20.70 −7.80 1.60 4.65 0.06 155.75
11 20 5 1 −498.1 41.00 −6.70 3.20 4.65 0.06 155.75
12 20 5 1 −462.4 24.20 −8.60 1.90 4.68 0.05 155.75
13 20 5 1 −498.8 24.40 −5.60 1.90 4.58 0.04 155.74
14 20 5 1 −461.9 20.50 −5.70 1.60 4.58 0.05 155.74
Innovation Methods for Residual Stress Determination …

15 20 5 1 −493.1 13.00 −6.00 1.00 4.60 0.06 155.75

16 45 5 5 −497.6 8.20 −10.90 1.90 5.15 0.10 155.71
17 45 5 5 −501.7 9.90 −12.50 2.00 5.14 0.09 155.70
18 45 5 5 −500.3 6.60 −11.40 1.30 5.13 0.09 155.71
19 45 5 4 −491.5 8.30 −9.70 1.60 4.95 0.09 155.69
20 45 5 4 −498 9.00 −12.70 1.80 4.94 0.07 155.73
21 45 5 4 −488.9 7.80 −10.90 1.50 4.95 0.08 155.70
22 45 5 3 −490.8 5.70 −8.50 1.10 4.74 0.05 155.73
23 45 5 3 −488 5.40 −9.10 1.10 4.74 0.05 155.73
24 45 5 3 −479.2 6.20 −7.80 1.20 4.74 0.06 155.72
(continued)
489
Table 1 (continued)
490

Location Var. No. Tilt max, ″ No. of tilt Spot size, mm r, Mpa (±) s, MPa (±) FWHM, ° (±) 2h, °
25 45 5 1 −469.9 11.70 −5.10 2.30 4.63 0.05 155.74
26 45 5 1 −474.6 11.70 −4.50 2.30 4.62 0.05 155.75
27 45 5 1 −470.3 11.30 −3.50 2.20 4.60 0.05 155.75
28 45 5 1 −494.8 8.30 −5.30 1.60 4.54 0.06 155.74
29 45 5 1 −485.7 7.20 −4.30 1.40 4.56 0.07 155.72
30 45 5 1 −485 6.00 −6.10 1.20 4.54 0.05 155.74
M. Sepsi et al.
Innovation Methods for Residual Stress Determination … 491

Fig. 2 Residual stress data in function of number of tilts

Fig. 3 FWHM data in function of number of tilts

The difference of the two locations (A and D location) can be explained based on
the results of Fig. 10. All the residual stress and FWHM data at A and D locations
are plotted. The spot size, the max tilting angle and the number of tilt are variables.
Three parallel measurements were performed with a given set of variables. The
Bragg angles have a small scatter in both locations due to the measuring parameters,
but the data of the two locations are markedly different. The A location represents
smaller Bragg angles and larger FWHM values. It can happen, if the soluted carbon
concentration is higher at the A location. The intensity difference of shot peening
492 M. Sepsi et al.

Fig. 4 Residual stress data in function of spot size

Fig. 5 FWHM data in function of spot size

can cause a shift of FWHM value, but in that case the Bragg angle is unaffected.
Because reporting stress results together the other parameters (FWHM, h) gives a
full description with much additional information is always highly recommended in
case of any component investigation.
Innovation Methods for Residual Stress Determination … 493

Fig. 6 Residual stress data in function of max tilt angle

Fig. 7 Residual stress data in function of spot size

The reproducibility of the measurement has been always the question in any case
of measurements. Figure 11 shows the reproducibility of stress measurement at the
D location in 9 different sets of measurements (a–i). Three parallel measurements
were carried out with a given set of variables. Now it is evident that the repro-
ducibility is very good, the scatter of the parallel measured data is smaller than the
error bar value.
494 M. Sepsi et al.

Fig. 8 FWHM data in function of max tilt angle

Fig. 9 FWHM data in function of spot size

Innovation Methods for Residual Stress Determination … 495

Fig. 10 All the Bragg angle and FWHM data at A and D locations, the spot size, the max tilting
angle and the number of tilt are variables. Three parallel measurements were done with a given set
of variables

Fig. 11 Reproducibility of stress measurement at the D location in 9 different sets of variables

(a–i). Three parallel measurements were done with a given set of variables
496 M. Sepsi et al.

4 Summary

The residual stress was measured non-destructively with a Stresstech Xstress 3000
G3R type centerless X-ray diffractometer on a carburized and shoot peened gear
wheel using the modified sin2w method. The data were recorded at two equivalent
parts of the gear in the perpendicular direction to the axis. The normal stress (r), the
shear stress (s), FWHM and Bragg angle (2h) were calculated at every measuring
time. To affect the applied measured parameters such as the spot size, the number of
tilting, and the tilting range were showed on based on the highest number of
measured data. The result can be summarized as follows:
• The two locations of the gear contain different stress and other results due to the
different solute carbon concentration and it is evident on based of simultaneous
investigation of normal stress, FWHM, and h.
• The increase of number of tilting has slightly effect on stress data. The stress
results of 11 tilts number shows only smaller error bar, but the difference is
much smaller than the difference of the result from the equivalent parts (location
A and D) of the gear. The FWHM data are unaffected by the number of tilting.
• The spot size effect on the stress data deviation is within the error bars, if even 6
times larger spot size was used, but the FWHM data are strongly increased as
the spot size is increasing. So the FWHM data can be comparable in case of
same spot size measurements only.
• The tilting range has the strongest effect on the stress data. In case of ±20°
tilting range the error bar can be 10 times larger than in case of ±45° which is
the usually used value. So, the sample cutting may have a smaller modification
effect on the original stress state and gives better results than the decrease of the
maximum tilting angle.
• The reproducibility was good enough in any case of a set of variables.

Acknowledgements The described article was carried out as part of the EFOP-3.6.1-16-00011
“Younger and Renewing University—Innovative Knowledge City—institutional development of
the University of Miskolc aiming at intelligent specialisation” project implemented in the
framework of the Szechenyi 2020 program. The realization of this project is supported by the
European Union, co-financed by the European Social Fund. The research work of Mate Sepsi was
supported through the New National Excellence Program of the Ministry of Human Capacities.

References

1. Handbook of residual stress and deformation of steel. ASM Int 2008, USA pp 347–358
2. Macherauch E, Hauk V (1987) Residual Stress Sci Technol 2:697
3. Soady KA, Mellor BG, Shackleton J, Morris A, Reed PAS (2011) The effect of shot peening on
notched low cycle fatigue. Mater Sci Eng, A 528(29):8579–8588
4. Cseh D, Mertinger V, Lukács J (2013) Residual stress evolution during fatigue test of a shoot
peened steel sample. Mater Sci Forum 752:95–104
Innovation Methods for Residual Stress Determination … 497

5. Withers PJ, Bhadeshia H (2001) Residual stress Part1—measurement techniques. Mater Sci
Technol 17:355–365
6. Schajer GS (2013) Practical residual stress measurement methods. Wiley, pp 140–161
7. In: Proceeding of 1st international conference on barkhausen noise and micromagnetic testing,
Sept 1–2, 1998, Hannove–Germany, Organized by Stresstech Group and IFW
8. Krawitz AD (2001) Introduction to diffraction in materials science and engineering. Wiley,
pp 119–143, 278–318
Author Index

A Filep, ádám, 483

Albrecht, Lajos, 123 Forrai, Mónika Kulcsárné, 257

B G
Balogh, András, 407, 469 Gáspár, Marcell, 453, 469
Bányai, Tamás, 245
Barna, Balázs, 123 H
Bartók, Roland, 323, 333, 383 Hegedűs, György, 109
Baumli, Péter, 187
Bencs, Péter, 39 I
Benke, Márton, 59, 187, 483 Illés, Béla, 245, 275, 341
Bereczky, ákos, 225
Béres, Gábor, 197 J
Béres, Miklós, 205 Jálics, Károly, 91
Bézi, Zoltán, 407 Jansen, Thomas, 439
Blága, Csaba, 355 Jármai, Károly, 13, 27
Bolló, Betti, 27
Borhy, István, 425 K
Bothfeld, Ralf, 439 Kárpáti, Viktor, 59
Bouzid, Ahmed, 333, 383 Kiss, Dániel, 235
Knopp, Ferenc, 99
C Koba, Máté, 323
Csáki, Tibor, 235 Korponai, János, 275
Cseh, Dávid, 483 Kovács, Gergely, 375
Cservenák, ákos, 69 Kovács, György, 391
Czap, László, 323, 333, 375 Kovács, László, 425
Kulcsár, Gyula, 257
D
Dobosy, ádám, 453 L
Dudás, László, 79 Lassú, Gábor, 135
Líska, János, 217
E L. Kiss, Márton, 383
Eggers, Jörg, 439 Lukács, János, 453, 469

F M
Farkas, József, 13 Majtényi, József, 59
Fekete, Tamás, 49 Mertinger, Valéria, 59, 483
Ferencsik, Viktória, 143 Mészáros, Ferenc, 123

© Springer International Publishing AG 2017 499

K. Jármai and B. Bolló (eds.), Vehicle and Automotive Engineering,
Lecture Notes in Mechanical Engineering, DOI 10.1007/978-3-319-51189-4
500 Author Index

Mezei, Imre János, 287, 309 T

Mihalik, Márk, 391 Takács, György, 3
Mikáczó, Viktória, 155 Tisza, Miklós, 197
Tokody, Daniel, 287, 309
N Tollár, Sándor, 39
Németh, Géza, 115 Török, Tamás I., 135
Tóth, ágota Bányainé, 275
P Tóth, Dániel, 3
Palotás, árpád Bence, 155, 171
Paripás, Béla, 205 V
Petrik, Máté, 27 Varga, Gyula, 143
Prém, László, 407 Varga, Zoltán, 391
Vásárhelyi, József, 333, 383
S Veres, Péter, 245
Sályi, Zsolt, 187 Veres, Zsolt, 187
Schuster, György, 287, 309
Sepsi, Máté, 483 W
Siménfalvi, Zoltán, 155, 171 Wagner, György, 341
Skapinyecz, Róbert, 341
Szabó, Szilárd, 39, 123 Z
Szabó, Tamás, 69 Zsemberi, Andor, 155, 171
Szepesi, Gábor, 27
Szilágyi, Attila, 3