ECAD/ECAE2004 - the Systems Realization Laboratory

ECAD/ECAE 2004 

PROCEEDINGS OF THE 1 ST INTERNATIONAL CONFERENCE ON ELECTRICAL / 

ELECTROMECHANICAL COMPUTER AIDED DESIGN & ENGINEERING 

Technical Papers 

University of Durham 

15 th & 16 th November 2004 

Edited by P. G. Maropoulos and D. Schaefer

© The University of Durham 2004 

All rights reserved. No reproduction, copy or transmission of this publication 

may be made without written permission. 

Except as otherwise permitted under the Copyright, Designs and Patents Act 

1988, this publication may only be reproduced, stored or transmitted, in any 

form or by any means, with the prior permission in writing of the publisher, or, in 

the case of reprographic reproduction, in accordance with the terms of a licence 

issued by a Copyright Licensing Agency. Enquires concerning reproduction 

outside those terms should be sent to the publisher. 

British Copyright Council 

http://www.britishcopyright.org.uk 

29-33 Berners Street, London, W1T 3AB. 

Published by 

The University of Durham, 

School of Engineering, 

South Road, 

Durham, 

DH1 3LE, UK. 

ISBN 0-9535558-3-6 

ii

ECAD/ECAE 2004 

PROCEEDINGS OF THE 1 ST INTERNATIONAL 

CONFERENCE ON ELECTRICAL / ELECTROMECHANICAL 

COMPUTER AIDED DESIGN & ENGINEERING 


15 th & 16 th November 2004 

Edited by P. G. Maropoulos and D. Schaefer 

Produced by D. G. Bramall, W.M. Cheung and C.D.W. Lomas 

iii

Contents about this CD 

Committees vi 

List of Sponsors vii 

About the Editors viii 

Editorial ix 

Part 1. Mechatronics 1 

Comos® ET/ME as System Solution in the Mechatronics Field 

M. F. Zaeh, F. Graetz and A. Mankel 

Vibration-Based Interface for Collaboration in Mechatronics Design 

T. Ito 

XML-based Product and Process Data Representation for Distributed Process 

Planning 

D. N. Šormaz, J. Arumugam and N. Neerukonda 

Part 2. Integration, Product Variants and Complexity 20 

The Integration of ECAE into the Complete Manufacturing Workflow (Presentation 

Only) 

T. Ward 

Release-Engineering – An Innovative Approach to Handle Complexity of Mechatronic 

Products 

G. Schuh, J.C. Desoi, M. Lenders, V. Witte 

Costing Issues Regarding Product Variant Design 

P. Baguley, D. Schaefer 

A New Method for Variant Design Technology in ECAD 

D. Schaefer 

Part 3. Micro Electro-Mechanical Systems (MEMS) 50 

Use of CoventorWare in the Design of innovative IC Probes Based on Microsystems 

Technology 

M.D.Cooke and D. Wood 

CAD Framework for the Design and Process Modelling of Micro-electromechanical 

systems (MEMS) 

B. Solano, D. Wood, S. Rolt and P.G. Maropoulos 

The Applicability of CoventorWare to RF MEMS 

A.J.Gallant and D. Wood 

Part 4. Process Automation 74 

Concurrent Electrical Engineering Methods 

N.S. Zughaid and P.D Hackney 

Electrical and Process Automation Framework for Engineering & Maintenance with 

3 rd CAE System Generation (Presentation Only) 

D. Mukherjee 

iv 

2 

6 

12 

21 

36 

41 

46 

51 

59 

67 

75 

80

Automatic Wiring in Switch Cabinets 

W. Vigerske, B. Stube, M. Plessow 

Part 5. PCB Design & Knowledge Management 94 

Genetic Algorithm for Xilinx-style FPGA Placement 

M. Yang, A. Almaini 

Maintaining Electromagnetic Compatibility (EMC) Through Design for Fabrication 

and Assembly of Printed Circuit Boards (PCB) 

T. Page, P.Baguley and D. Schaefer 

Ensuring Electromagnetic Compliance in Printed Circuit Boards through Design and 

Assembly Guidelines 

T. Page, P. Baguley and D. Schaefer 

Knowledge Transfer in Networks of Competence 

C. A. Schumann, C. Tittmann and K. Grebenstein 

Part 6. Late Papers 116 

Fuzzy Sliding Mode Control Solution to an Inverted Pendulum: An Implementation 

Issue 

M. Bouchoucha, A. Souissi and A. Derdouche 

Modelling of Transport Flows for Inter-disciplinary Systems Integration (Abstract 

Only) 

A. Suslenkovs and A. Levchenkov 

Modelling of Power Distribution Systems with the help of Internet Based Data 

Processing Technologies (Abstract Only) 

D. Rihtere and A. Levchenkov 

Author Index 125 

Searchable Index 127 

v 

90 

95 

101 

106 

111 

117 

123 

124

International Scientific Advisory Committee 

Prof. P.G. Maropoulos (Chair) 

University of Durham, UK 

Dr. M. Plessow 

GFaI e.V., Berlin, Germany 

Prof. P. Xirouchakis 

Swiss Federal Institute of Technology, 

Switzerland 

Prof. W.F. Reiter 

Oregon State University, USA 

Prof. A. Bernard 

Ecole Centrale de Nantes, France 

Prof. G. Schuh 

RWTH-Aachen, Germany 

Prof. K. Feldmann 

University of Erlangen-Nuremberg, Germany 

Prof. D. Qin 

Chongqing University, China 

Prof. K. Cheng 

Leeds Metropolitan University, UK 

Local Organising Committee 

Prof. P.G. Maropoulos Dr. D. Schaefer 

Dr. G. Coates Dr. H. Long 

Dr. Q. Wang Dr. P. Matthews 

Mr. O. Vogt Mr. C.D.W. Lomas 

Mr. P. Baguley Mr. B. C. Rogers 

Mr. P. Chapman Mr. W.M. Cheung 

Mr. D. G. Bramall 

vi 

Dr. D. Schaefer (Co-Chair) 

University of Durham, UK 

Prof. D. Roller 

University of Stuttgart, Germany 

Prof. T.R. Kurfess 

Georgia Tech, Atlanta, USA 

Prof. J. Duflou 

K.U. Leuven, Belgium 

Prof. D. Brissaud 

University of Grenoble, France 

Prof. P. Cunha 

IST, Portugal 

Dr. Steve Newman 

Loughborough University, UK 

Dr. R.D. Sriram 

NIST, USA

List of Sponsors 

Technische Computer 

Systeme, Suessen 

GmbH 

ePLAN Rittal Ltd 

Euro Info Centres 

(ECI) 

One North East 

Gesellschaft fuer 

Informatik e.V. 

International Journal 

of Computer 

Integrated 

manufacturing 

Global Digital 

Enterprise Research 

Laboratory 

Agility 

Derwentside 

Engineering Forum 

NetPark 


vii 

http://www.tcs-s.de/english 

http://www.eplan.co.uk 

http://www.euro-info.org.uk 

http://www.onenortheast.co.uk 

http://www.giev.de/allgemeines/indexenglish.htm 

http://www.tanf.co.uk/journals/titles 

/0951192X.asp 

http://gderl.dur.ac.uk/ 

http://www.dur.ac.uk/agility/ 

http://www.def.org.uk/ 

http://www.uknetpark.com/ 

http://www.dur.ac.uk/ 

All links verified on 26 th October 2004

About the Editors 

Paul Maropoulos 

Dipl Ing, MSc, PhD, Eur Ing 

Paul Maropoulos is Professor of Engineering in the School of 

Engineering at the University of Durham. He directs the 

research of the Design and Manufacturing Research Group 

and is Joint Director of the Centre for Industrial Automation 

and Manufacture. He has close links with industry and heads 

the ‘Agility Project’ - a manufacturing consultancy 

specialising in lean methodologies. 

His first degree was in Mechanical Engineering from the University of 

Thessaloniki in Greece and he subsequently obtained his MSc and PhD 

degrees from the University of Manchester Institute of Science & Technology 

(UMIST). 

He is actively pursuing research in the following areas; design and 

manufacturing integration, collaborative and distributed product and process 

development, technical evaluation of the supply chain, new generation process 

planning and routing using aggregate methods, aggregate process modelling, 

assembly planning and system design, distributed tool selection and support for 

machining, factory design and cell formation, capability analysis and laser 

metrology for manufacturing systems. 

Dirk Schaefer 

Toolmaker, BSc, MSc, PhD 

Dirk Schaefer is a Lecturer in the School of Engineering at 

the University of Durham. Before this, he was a Research & 

Teaching Assistant at the University of Stuttgart, Germany, 

where he earned his PhD in Computer Science. During this 

time, Dirk also held part-time positions as an Assistance 

Professor in Computer Science at the University of Applied 

Sciences at Aalen, and as a Lecturer in Computer Science at 

two private IT Academies at Esslingen and Dresden. Furthermore, he was the 

Managing Director of an IT consulting firm, which he founded in 1999. 

Prior to working in academia, he gained experience as a Software Engineer with 

HOCHTIEF GmbH in Essen, Germany. In this role he designed and 

implemented software modules for an object-oriented Civil Engineering CAD 

system. Dirk started his career as an apprentice Toolmaker with Graebener 

Press Systems, one of Germany’s leading metal forming companies, where he 

specialised on CNC machining and the manufacture of compound tool sets for 

knuckle joint presses. Between working as a Software Engineer and Toolmaker, 

he obtained his first degree in Mechanical Engineering at the Technical College 

in Siegen, Germany, and graduated some years later with a Masters degree in 

Mathematics from the University of Duisburg, Germany. 

His current research interests are focused on the development of novel 

approaches and computer applications involved in the interfaces between 

applied Computer Science, Mathematics and Engineering. Recent work has 

primarily addressed several aspects of CAE and CAD/CAM technology in 

mechanical, electrical and electromechanical engineering. 

viii

Editorial 

Industry is undergoing a rapid transformation internationally, characterised by the 

liberalization of global markets, increased mobility of people and capital and the constant 

requirement for exploiting the enabling capability of the internet to rapidly add value to 

products and services. Increased competition has also led to the gradual consolidation of 

sectors as companies adopt corporate alliances and mergers in order to manage their 

spiralling R&D and marketing costs across different markets. Currently, the defence and 

aerospace sectors are largely consolidated on both sides of the Atlantic and the same is true 

for the automotive, pharmaceuticals and chemicals industries. 

The removal of protectionist policies, coupled with the availability of a skilled workforce and 

comprehensive communication technologies in many parts of the world, has allowed large 

enterprises to adopt globalisation and plan manufacturing and design operations on a truly 

global basis. The benefit of this approach is that the resources of an enterprise can be finetuned 

to allow competitive operations in traditional major markets, like the North American 

and the European regions, as well as covering the varied demands of emerging new markets. 

In addition, an enterprise can seek to exploit specific scientific skills that may be available in 

abundance in certain countries, such as design technology and mathematical analysis in 

Russia and software engineering skills in India. Globalisation can be rewarding for industry at 

an enterprise level, but it can be problematic for small and medium size enterprises (SMEs), 

which underpin the operation of larger enterprises and are crucial for the economic growth of 

specific countries and regions. Initial fears of a massive employment drift to low wage regions 

have largely not been realized. A good example of the opposite trend is the recent growth of 

manufacturing industry in the United States, which was principally sustained by developing 

business practices that apply computer-based technologies and use a first class electronic 

networking and communications infrastructure. Clearly, computer-based methods for design 

and manufacture supported by electronic networking and communication infrastructure (ecommerce) 

represent an essential level of technology required by SMEs in order to support 

competitive manufacturing operations on a regional and global basis. 

Globalisation and enterprise consolidation amplify the effects of socio-economic events 

and market requirements in industries and economies of the industrialised world, in much 

more direct and immediate ways than previously. These conditions create the need for: 

(i) Highly agile and innovative corporations that can seize and exploit market 

opportunities, world-wide. 

(ii) Effective management of corporate revenue streams across major technology 

development phases. 

(iii) Formation of dynamic supply networks and development of flexible manufacturing 

capabilities within the extended enterprise, based on the sound management of 

the product’s life cycle. 

Industry and research organizations can now capitalize on recent advances in computer 

modelling, graphic visualization and distributed information management, in order to positively 

impact product development and realization and develop objective risk mitigation strategies 

on a global basis. These technologies will allow the development of new business processes 

to take advantage of the rapid formation of production networks and effectively manage 

workflow during all phases of a product’s life cycle. “Digital Enterprise Technology” (DET) 

represents this emerging, new synthesis of technologies and systems for product and process 

development and life cycle management on a global basis. DET can be defined as “the 

collection of systems and methods for the digital modelling of the global product 

development and realization process, in the context of lifecycle management”. 

DET is implemented by the synthesis of technologies and systems from five main technical 

areas, the DET “cornerstones”, corresponding to; (i) distributed and collaborative design, (ii) 

process modelling and process planning, (iii) production equipment and factory modelling, (iv) 

digital to physical environment integrators and (v) enterprise integration technologies. The first 

ix

three technical areas of DET can now be performed entirely in the digital domain, utilising the 

enhanced graphics and computer processing technologies developed over the past five years 

as well as the communication infrastructure of the internet. The fourth category is of particular 

importance as it involves methods for the bi-directional integration of the digital and physical 

environments in order to achieve risk mitigation at product and system levels using shop floor 

based metrology and discrete event simulation respectively. The fifth area of DET includes 

methods that are predominantly employed to manage physical resources when these have 

been released for manufacturing and as such are used when product realization has 

commenced. The fifth area also includes web-centric methods for product data management 

that is applicable from the very early stages of product development and spans the five areas 

of DET. 

One particular area of Computer Aided Engineering on which the framing Digital Enterprise 

Technology (DET) philosophy may be applied to is the development of Computer-Aided 

Design and Engineering systems and software tools for electrical/electromechanical 

engineering (ECAD/ECAE). These systems have already undergone major changes during 

the past couple of years and are still subject to groundbreaking change and innovation in 

respect to both systems technology and philosophy. 

The early, so-called 2 nd generation ECAD systems, which were hardly more than a digital 

replacement of the drawing board, have recently been replaced by an entirely new 

development of 3 rd generation ECAD/ECAE systems. These new systems are called 

Electrical Engineering Solutions (EES) and focus on supporting an entirely integrated 

engineering process from design to manufacture. One of their major philosophies is to 

facilitate global concurrent engineering along the whole product lifecycle chain. Specifically, 

the integration of electrical and mechanical engineering tools is of increasing relevance, in 

order to allow interdisciplinary, bi-directional product modelling, as well as process planning. 

Hence, the development of appropriate interfaces and international standardisation issues 

has become an essential task. 

Current R&D activities within the area of ECAD/ECAE mainly focus on: 

1. ECAD/ECAE Product and Process Modelling Automation 

- Product modelling and product data management 

- Process modelling and workflow management 

- Agile design, configuration and manufacture of product variants 

- Automatic generation of schematics and project documentation 

2. Distributed and Collaborative Engineering 

- Integration of electrical with mechanical engineering systems 

- Representation of interdisciplinary design knowledge 

- Modelling and propagation of multidisciplinary constraints 

- Multidisciplinary global concurrent engineering 

3. ECAD/ECAE Product Platforms and System Development 

- Novel architectures for ECAD/ECAE systems 

- Current state-of-the-art-systems: opportunities and challenges 

- Advances in database technology for engineering applications 

- Emerging trends for developing future ECAD/ECAE software 

4. International ECAD/ECAE Standardisation Issues 

- Standardised exchange of product model data (STEP – ISO 10303) 

- Embedding of international standards into ECAD/ECAE software 

- Standards for interdisciplinary systems integration 

- Standardised internet based data processing 

5. Implementation of ECAD/ECAE Systems in Industry 

x

- Business process re-engineering 

- Validation and selection criteria for ECAD/ECAE systems 

- Implementation/Migration strategies and case studies 

- Integration of contemporary systems into legacy environments 

6. Beyond the Borders of Classical ECAD/ECAE 

- ECAD/ECAE systems in hydraulics and pneumatics 

- Design and development of mechatronic systems 

- Simulation of mechatronic systems 

- From mechatronic to micro-electro-mechanical-systems 

This Proceedings volume contains papers submitted for the 1st International Conference on 

Electrical/Electromechanical Computer Aided Design and Engineering (ECAD/ECAE’04). 

This meeting is the first international event on ECAD/ECAE Technology and follows three 

national meetings consecutively held at Stuttgart University, Germany, between 2000 and 

2003. 

The main aim of ECAD/ECAE’04 is to provide an international forum for the exchange of 

scientific knowledge and industrial experience regarding state-of-the-art in the various 

aspects of Electrical/Electromechanical Computer Aided Design and Engineering. 

Furthermore, ECAD/ECAE’04 aims to increase the international awareness of a rapidly 

growing field of engineering research and systems development. 

The Proceedings are organized in 7 main sections, which contain some 20 papers covering a 

wide area of ECAD/ECAE related research, from product modelling and internet based data 

processing; automation and systems integration; to mechatronics and micro electromechanical 

systems (MEMS). Papers are organised as presented during the various sessions 

of the conference. 

ECAD/ECAE’04 has special significance for the University of Durham and the North East 

region, since it is the third major meeting (following CAPE’99 and DET’02) in production 

engineering to be organized and hosted by Durham University. The North East of England is 

a region with a long tradition in manufacturing and engineering. The future development of the 

region, and in particular of County Durham, is based on the adoption and exploitation of new, 

knowledge intensive technologies that can give local companies a competitive advantage on 

a global basis. Hence ECAD/ECAE, within the framing concept of DET, is thematically ideally 

aligned with the strategic development plans of the North East region and will benefit local 

industry and regional agencies of economic and industrial development. 

The Editors would like to gratefully acknowledge the help and support of all those who helped 

in any way during the months of preparations for ECAD/ECAE’04. In particular, sincere 

thanks go to the ECAD/ECAE’04 Local Organising Committee Members, who have played a 

major role in all aspects of ECAD/ECAE’04 organisation and management. The Editors also 

wish to acknowledge the support of the International Advisory Committee, the financial 

support of the major ECAD/ECAE’04 sponsors; TCS GmbH (Germany) who produce the 

promis engine e® / SIGRAPH CAE software, OneNorthEast, Eplan/Rittal Ltd. 

Professor P.G. Maropoulos 


Editor and ECAD/ECAE’04 Chairman 

Durham, November 2004 

xi 

Dr D. Schaefer 


Co-Editor, ECAD/ECAE’04 Co-Chairman 

and Manager

PART 1 

MECHATRONICS 

1

Comos ® ET/ME as System Solution in the Mechatronics Field 

Michael F. Zaeh (iwb), Florian Graetz (iwb), Alexander Mankel (innotec) 

Institute for Machine Tools and Industrial Management (iwb), Technische Universitaet Muenchen, Garching, Germany 

Abstract: 

innotec GmbH, Schwelm, Germany 

This article describes the reasons and the aims of the cooperation between innotec GmbH, a vendor for 

CAE-systems, and the Institute for Machine Tools and Industrial Management (iwb) of the Technische 

Universitaet Muenchen. This collaboration resulted in the COMOS ® ME software, a system solution for the 

development of mechatronic products. A short overview over the system concepts and possible fields of 

applications is given. 

Keywords: 

Computer Aided Engineering, Simultaneous Engineering, Mechatronic System Design 

1. INTRODUCTION 

The term Mechatronics was formed in Japan in 1969 by 

the president of Yaskawa Electric Corp., Ko Kikuch [1]. 

This expression was made up of the words "Mechanisms" 

and "Electronics" and was registered as trade name from 

1971 to 1982. With the increasing capabilities and 

declining costs of microprocessors, the importance of 

software in industrial and consumer applications grew 

gradually. Nowadays, Mechatronics is commonly 

understood as the greatest possible integration of 

mechanics, electrics and software to one product [2]. 

With time, the number of aspects, which have to be 

reflected during the development process for a product, 

grew tremendously. Therefore, the expertise of new fields 

of engineering – such as fluid or software engineers – 

became essential. To enable developers to complete their 

steadily complicating tasks within a limited time frame, 

new approaches and methods were required. The classic 

procedure for mechanical engineering as described in [3], 

was later enriched with aspects relevant for electrical 

engineering and led to new design guidelines, for example 

[4]. With the growing importance of software components 

in mechatronic systems and shortening development 

times, highly specific methods for certain industries have 

been evolved. One example for machine tools is given in 

[5]. 

2. BACKGROUND 

As for many fields of applications, different approaches 

have been set up, the methods for mechatronic 

engineering diverged. Hence, unification became 

desirable and was accomplished by the Association of 

German Engineers (VDI) with guideline no. 2206 "Design 

Methodology for Mechatronic Systems" [6]. It contains 

recommendations of tools covering the aspects of 

mechatronics. Among them are tools for 

• requirement engineering, 

• computer aided design (CAD), 

• finite element methods (FEM), 

• boundary element methods (BEM), 

• multi-body simulation, 

• computational fluid dynamics (CFD), 

• control engineering, 

• computer aided engineering (CAE), 

• computer aided software engineering (CASE), 

• hardware-in-the-loop (HIL), 

2 

• software-in-the-loop (SIL), and 

• product data management (PDM). 

Although [6] is a binding design guideline and covers the 

needs of mechatronic engineering, it has to be specifically 

accommodated for every application. 

As VDI guideline 2206 is tailored for individualized 

mechatronic mass products, it hardly matches the 

requirements for the development of mechatronic 

investment goods, since in most cases they are produced 

at a quantity of or nearby one. 

3. REQUIREMENTS 

The development at most manufacturers for mechatronic 

systems today lacks of communication especially between 

the mechanical and electrical design. This situation forces 

every faculty to await the completion of the design for all 

preliminary phases. Therefore, the engineering of 

machines can best be described as sequential [7]. 

3.1 Method 

The basic concept of the approach suggested in this 

paper is shown in figure 1. To shorten the time of 

development, a pre-design phase is set up, which covers 

most aspects of interdisciplinary communication. 

Therefore, the mechanical, electrical and software design 

can run in parallel, leading to a shorter time span between 

the purchase order for a machine and its delivery. 

Due to the alikeness of fluid and electric engineering, at 

least on an abstract level, the same methodology is 

suggested for both fields of design. In consequence, 

benefits resulting from a general concept can be achieved 

[8]. 

Figure 1: Concept of the Method supported by Comos ®

3.2 Tool 

To allow the advantages of a methodology to pay off in 

industrial application, its implementation in software is 

mandatory. To achieve the user's acceptance, 

extraordinary improvements over conventional tools have 

to be realized. The main goals during the project 

described in this paper were 

• to guarantee data consistency, 

• to automatically generate key documents, and 

• to include the predevelopment phase in a tool used 

for detail engineering. 

4. APPROACH 

Based on the requirements explained in chapter 3, a 

method and a tool supporting it, have been developed. 

4.1 Method 

Figure 2 shows steps to be taken during the pre-design 

phase (also see 3.1), as suggested in [5]. It is divided into 

the functional description of a machine and its 

infrastructure planning as described below. 

Functional Description 

A machine is structured according to functional aspects as 

recommended in [9]. For every function, an operational 

principle (figure 3) and a sequence as a project network 

[10] are drawn. Each step in this project network is 

assigned to a degree of freedom. For every degree of 

freedom a state graph including all attainable states is 

defined (also see [4], [11]). Temporary and final states 

can be distinguished. While final states are determined by 

constant energy within the system of interest, temporary 

states are required to change from one final state to 

another. For every transition from a temporary to a final 

state, the amount of energy within the system has to be 

determined. Therefore, required measurement quantities 

for every degree of freedom can be derived. One example 

of a state graph for rising and lowering the upper die is 

given in figure 4. 

Figure 2: Engineering as Supported in Comos ® ME 

Figure 3: Operating Principle for a Press 

3 

Infrastructure Planning 

Figure 4: Simple State Graph 

After the functional description is completed, the concepts 

defined in operation principles, network diagrams and 

state graphs have to be realized with hard- and software 

components. Therefore, each degree of freedom is 

assigned to an actuator and for every measurement 

quantity a sensor is chosen. As suggested in [8], setting 

elements – for example valves or relays – elements are 

added as required to every degree of freedom. Figure 5 

illustrates that then the electrical input and output signals 

in the machine field are determined. Therefore, the 

signals for the CNC cabinet can be obtained (also see 

figure 6). 

Figure 5: Actuators, Sensors and Setting Elements 

Figure 6: Connections for Field and Control Components

Derived Documents 

As shown in [12], sequential function charts according to 

[13] can be received from state graphs and network 

diagrams. This kind of document contains a considerable 

amount of valuable information for the electrical and fluid 

design of a machine tool. However, as it does not support 

the mechanical engineering process, it is rarely provided. 

Therefore, a method of automatically generating 

functional charts improves the interdisciplinary 

communication. 

Furthermore, I/O lists can be easily created from 

connection diagrams as shown in figure 6 (also see [12]). 

4.2 Tool 

The documents and data object described in section 4.1 

have been integrated into the software platform Comos® 

[5], a product, which is commonly used in chemical 

engineering, which is – like mechatronic engineering – 

highly interdisciplinary. 

Macros have been provided to perform the tasks of 

automatically generating key documents, such as 

sequential function charts or I/O-Lists. The architecture of 

Comos ® is illustrated in figure 7. The component object 

server provides an interface between a database, which 

saves all project information, and every single engineering 

application. Since every piece of information is stored 

once in the database and depending views link to one 

unique record, data consistency is ensured. 

The engineering applications "Basic Logical Diagram" 

developed by the iwb and "Electrical & Instrumentation" 

as well as "Hydraulics & Pneumatics" from innotec have 

been combined to the Comos ® ME bundle. 

Figure 7: Architecture of the Comos ® Product Family 

4 

5. CASE STUDY 

To prove the feasibility of the method and tool portrayed in 

chapter 4, a complete milling machine has been mapped 

in Comos ® in cooperation with an international vendor for 

machine tools [14]. All actuators and sensors totaling to 

roughly 350 I/O-signals have been modeled in an Access ® 

database. Figure 8 shows the machine and a screenshot 

of a document describing parts the electrical wiring. 

Information about actuators, sensors and cables have has 

been exported to Pro/Engineer via an XML interface 

considered in [12] (figure 9). Furthermore, raw concepts 

for the PLC software have been generated from the 

sequential function charts and imported into Simatic 

Manager in an IEC 61131 format (figure 10). 

Figure 8: Electrical Connections of a Machine in Comos ® 

Figure 9: Interface to 3D-CAD-System 

Figure 10: Interface to PLC-Environment

6. CONCLUSIONS AND FUTURE WORK 

The development of the Comos ® ME system is an 

example, how the cooperation between a university and a 

software vendor can lead to innovative and promising 

products. The research work completed by the iwb and 

the software framework from innotec GmbH lead to a new 

generation of engineering tools, ready to enter the market. 

The viability the method and the toll described in chapter 

4 has been proven in a case study sketched in chapter 5. 

Further work will focus on customizing the COMOS ® ME 

to the requirements of users. The integration of the 

engineering environment and certain simulation tools, for 

example hardware-in-the-loop systems as described in 

[15] and [16], may lead to even more improved 

development cycles in the field of mechatronic 

engineering. 

[1] Harashima, F.; Tomizuka, M.; Fukuda, T.; 1996: 

Mechatronics – What Is It, Why, and How? – An 

Editorial, IEEE/ASME Transactions on Mechatronics, 

Vol. 1, No. 1, 1996, pp. 1-4 

[2] Reinhart, G.; Egermeier, H.; Thieke, S.; Dürrschmidt, 

S.; 2001: Mechatronik, In: offprint of Enzyklopädie 

Naturwissenschaft und Technik, addendum 7 

[3] VDI 2221, 1993: Methodik zum Entwickeln und 

Konstruieren technischer Systeme und Produkte. 

[4] VDI/VDE 2422; 2001: Entwicklungsmethodik für 

Geräte mit Steuerung durch Mikroelektronik. 

[5] Zaeh, M. F.; Graetz, F.; Rashidy, H., 2003: An 

Approach to Simultaneous Development in Machine 

Tool industry, Proceedings on International 

Workshop on Modeling and Applied Simulation, pp. 

128-133, Bergeggi, Italy. 

[6] VDI 2206, 2004: Design Methodology for 

Mechatronic Systems. 

[7] Zaeh, M. F.; Graetz, F.; Rashidy, H., 2004: 

Installation Planning for Mechatronic Production 

Systems, Proceedings on the IEEE-Conference 

"Mechatronics & Robotics 2004", pp. 315-319, 

Aachen, Germany. 

[8] Zaeh, M. F.; Graetz, F.; Gerdelmann, U.; 2003: Neue 

Konstruktion – Mechatronische Systeme Systeme 

erfordern neue Konstruktionswerkzeuge, SPS- 

Magazin, vol. 11//2003, pp. 68-70. 

[9] VDI 2800, 2000: Value Analysis. 

[10] DIN 69900-2, 1987: Project Controlling, Project 

Network Techniques, Methods of Presentation. 

[11] PG3, 2001: Projektgruppe 3, Richtlinie 

Funktionsbeschreibung. 

[12] Zaeh, M. F.; Graetz, F.; Rashidy, H., 2004: 

Installationsplanung – Entwicklung von 

mechatronischen Produktionssystemen. In: 

Konstruktion 7/8-2004, pp. 59-62. 

[13] IEC 60848, 2002: GRAFCET specification language 

for sequentional function charts. 

[14] Zaeh, M. F.; Graetz, F.; Rashidy, H., 2004: A 

Platform for Installation Planning – a case study. In: 

Proceedings of the CIRP Design Seminar 2004, 

Cairo, Egypt. 

[15] Maier, M.; Dierssen, S.; Bathelt, J., 2004: Die 

Virtuelle Maschine – oder wie Inbetriebnahme vor 

Montage möglich ist. In: Seminarband zum 

Chemnitzer Produktionstechnischen Kolloquium 

2004, pp. 365-380. 

5 

[16] Zaeh, M. F.; Poernbacher, C.; Wuensch, G., 2003: 

Emerging Virtual Machine Tools. In: Proceedings of 

the DETC 2003, 29 th Design Automation 

Conference, Chicago, Illinois.

vibration-based interface as one sort of the forcefeedback 

one. 

Figure 6: System configuration of vibration-based 

interface. 

Figure 6 shows a configuration of vibration-based 

interface developed in this reseach. 4 sets of vibration 

motor are used to give a vibration siglal to the user. The 

vibration motors are controlled by PIC microchip on a 

control circuit, based on the signals transmitted from PC 

through RS-232C serial line. This interface is used by 

hand movement of a user, of which position in 3D space 

is traced by a magnetic sensor system composed of a 

controller, transmitter, and receiver. 

Figure 7: User interaction with vibration-based interface. 

Figure 7 shows a user interaction with a virtual object via 

this vibration-based interface. The user wearing a glove 

and a headband in Figure 7 is under manipulation on the 

object in the screen. The glove is equipped with magnetic 

sensor to trace the position and orientation of hand to 

manipulate the object which is shown in the computer 

screen. The headband is equpped with 4 vibration motors, 

which start and/or stop vibration in various strength 

corresponding to the signals to feed back to the user. 

When the user touches the object, the system returns a 

signal so that the user can feel that his hand touches the 

object. The vibration interface is equipped with 4 vibration 

motors. With the combination of these 4 motors, the 

system provides the direction toward which the user 

touches the object as well. If the user touches from the 45 

degree to the object front, vibration motor 2 and 3 are on 

in the same strength of vibration, whereas, for example, 

the user touches the object from 60 degree to the object 

front, or more horizontal direction, vibration of motor 3 is 

stronger than that of 2. If the user touches around the 

object, the vibration moves around the object according to 

the touching point, so that the user can feel from which 

direction the user is currently touching the object as a real 

time interaction. 

Figure 8: Vibration mechanism in remote object 

manipulation. 

Figure 8 shows a series of interaction with virtual object 

with virtual hand which is controlled by the user’s hand 

through the vibration-based interface. When the hand is 

away from the object as show in Picture 1 in Figure 8, no 

vibration is given to the user. The user moves around the 

hand and it touches to the object from the front direction 

as shonw in Picture 2, vibration starts on motor #2 on the 

front position. During the move of object by pushing with 

hand, vibration becomes stronger so that the user has a 

feeling of pushing the object. The user keeps on pushing 

the object along the crank-shape route, changing the 

direction of pushing from front-foward to right-forward, 

then front-forward direction, during which the vibration site 

changes corresponding to the move so that the user also 

9

has a feeling of pushing to each direction to follow the 

route to reach the goal. The pictures on the right hand 

side in Figure 8 shows how the vibration position changes 

during this operation. 

Figure 9: Snapshot of computer screen for vibrationbased 

interface. 

This vibration-based interface not only provides a control 

function with digital objects but also provides an interface 

to control a mechatronics object. For example, Figure 9 

shows a snapshot of teleoperation interaction to the 

remote object. The user in the local site is under video 

conference with a remote site, where a cylinder-shaped 

object can be recognized through video Window 1 on the 

left side of the screen. What the local user is doing is to 

move the object with his hand. The object is equipped 

with 2 DC motors so that it can be controlled to move in 

any direction if a control signal is given to it. A hand in 

Window 2 on the right side of the cylinder-shaped object 

is a virtual hand which is corresponding to the real hand of 

the manipulating person in terms of movement of hand. 

The user can feel the touching point on the object, and 

also feel the vibration indicating the direction to which the 

user should move it. It also means that the user can move 

the object to the appropriate direction even if the user 

cannot see the window, because the vibration signal gives 

the direction to him. The participants in the video 

conference can manipulate the object with a touch and 

feel of movement, which actually controls the real object. 

3.2 Force-feedback lever interface for mechatronics 

design review 

Physical interaction plays a very important role in design 

review of mechatronics products, for example, a 

mechatronics devive. Designers study not only the 

appearance of the device but also embedded function and 

usability of it. Designers touch the device, use it and 

check the function to discusses the feasibility of the 

design. Design review could also be carried out in a video 

conference system over the network so that participants 

remotely distributed should join the discussion. However, 

physical interaction with the device is not available in a 

conventional video conference system, including 

CELAVIS. This research pays attention on the idea of 

physical interaction in video conference system, for 

example, the usage of vibration-based interface as 

described in the previous section. This section describes 

a senario of physical interaction for design review of a 

mechatronics device. 

Figure 10 shows a force-feedback controller which is 

under development as a part of this research. The lever 

on the right side of the box is designed to control the up- 

10 

down movement of a target device. A DC motor in the box 

generates a dynamic torque based on the angle of 

movement detected by potentiometers. The lever returns 

a force-feedback to a user so that it can be used as an 

interface for object manipulation. 

Figure 10: A controller with a force-feedback lever. 

Figure 11 shows a snapshot of computer screen during 

video conference to have design review on some devices 

in dental applications. Window 1 in Figure 11 shows a 

physical measurement device, called micro-surveyor, 

which is used to examine abutment teeth, design dentures, 

and to make parallel measurements of abutment teeth for 

bridges in dental applications. The window shows a 

measurement to check the jaw bone condition using a 

physical jaw bone model. The pen of the micro-surveyor 

has free movement on x-y horizontal surface, and updown 

movement on its vertical axis. 

Window 2 in Figure 11 shows a digital prototype which is 

identical to the micro-surveyor and jaw bone in Window 1. 

We have implemented manipulation software so that this 

digital compass can be manipulated by a controller such 

as joystick. A user can move up-and-down the pen to 

check the jaw condition, however, the user cannot have a 

feeling of touching to the jaw bone even if the pen touches 

it or penetrate to inside of the jaw bone. We have adopted 

the force-feedback lever controller shown in Figure 10 to 

manipulate the movement of the pen. In combination with 

another controller under development, the user can 

control the movement of a virtual micro-surveyor in a 

touch and feeling manner, which could control an actual 

mechatronics micro-surveyor if it is designed to be 

controllable by PC. 

Figure 11: Snapshot of computer screen of interactive 

manipulation in video conference.

4. Concluding remarks 

A number of commercial or free systems are available, as 

the increasing demand on video conference for business, 

technical, and routine meetings. The main objective of 

these systems is to provide an environment for 

audio/video communication for participants. If we need to 

have a video phone conversation with the counterparts, 

then it surely works fine. We have developed CALAVIS 

system as we described in this paper, and we have 

confirmed that our small group meeting which we held a 

couple of times between Japan and US sites works quite 

fine. 

In our video meeting experiments, we also have found 

that when we have a face-to-face meeting, what we would 

like to share is not only the information which is discussed 

in the conference but also the data, software, or physical 

devices on which the agenda of meeting might be related. 

Currently available system including CALAVIS does not 

provide such an environment where participants can 

share those items. What we are pursuing as one of the 

objectives in this study is to setup an environment where 

participants can share a virtual workspace where they can 

work with the common sharing of those items as well as 

have a video conference. 

This study focuses on the physical aspects of interaction 

in video conference, which means that participants not 

only should share information/data through visual 

interactions but also should share them through physical 

interactions to work together much more collaboratively. 

For this objective, this paper presented a web-based 

video conference system with vibration-based interface to 

manipulate a remote mechatronics object through virtual 

space. Our system makes it possible to have physical 

interaction with the object over the network. 

The vibration-based interface presented in this paper is 

not as powerful as forced-feed back device interaction but 

users can feel the manipulation. While vibration signal is 

commonly used to draw attention from users, such as 

silent mode of mobile phone, and to let them know 

something, such as phone call, or new messages, etc., 

but it is not suitable to guide physical direction to users to 

support manipulation. With the combination of multiple 

vibration motor based on a simple protocol to show 

direction of movement, the interface of our system 

provides not only signals of notice, but also guidance of 

manipulation. For example, a user may receive vibration 

signal at the time of contact with an object during 

manipulation, and then may be given directional vibration 

to move accordingly. Reporting the result of experiment 

using a simple mechatronics prototype via vibration-based 

interface for collaboration, the paper discussed the 

feasibility of the interface for collaboration as well as the 

usability of the interface. 

BOCOLLA interface presented in this paper is also a new 

approach to support collaborative work over video 

conference. When we have a meeting, we often use a 

whiteboard-like panel to write down words, phrases, 

figures, marks, etc., which surely support understanding 

each other with reducing difficulty and mistakes caused by 

miscommunication, and with increasing easiness and 

mutual understanding supported by visual aides. With 

only a marker pen, participants can not only control 

computers located locally or remotely, but also even 

manipulate teleoperation to control a remote machine 

located on a remote site. Although what we did in this 

study was limited to a simple operation on a miniature 

crane, but we are actually able to manipulate various 

kinds of devices, including remote computers, mechanical 

machines, electrical devices, home appliances, etc. only if 

we can make it available to control them via IT network, 

which can be available in a variety of ways. 

What we have employed in our test environment was a 

teleoperation as an example of collaborative work during 

video conference. To do so, we have prepared a toy 

crane and controller, both of which can be connected to 

PC either locally or remotely, so that an operator can 

control it during video conference. BOCOLLA interface 

allows participants to control the crane during the 

conference. Vibration-based interface presented in this 

paper can also be used as an interface. Participants can 

have a touch-and-feel of control on the crane. The forcefeedback 

lever presented in this paper can also be 

integrated in this test environment to support physical 

interaction in video conference. 

The idea of physical interaction presented in this paper is 

to support much more collaboration in mechatronics 

design over the network. We will continue our research, 

and work much more on the development on each 

interface modules and subsystems to implement an 

integrated test environment. Using such an integrated 

environment, we will study the feasibility of our idea in 

more practical scenarios. 

11 

Acknowledgement 

This research was supported in part by the Grants-in-Aid 

for Scientific Research in 2003-2004 (Scientific Research 

(C) 15500347) from the Japan Society of Promotion of 

Science. The author would like to thank Prof. Teisuke 

Sato of Department of Mechanical Engineering, and Prof. 

Tetsuo Ichikawa, Dr. Yoritoki Tomotake, DDS. Keiko 

Shibutani of School of Dentistry of the University of 

Tokushima, Prof.Koichi Sairyo of School of Medicine at 

the University of Tokushima, Dr. Andrew Milne of Stanford 

University. The author would also like to thank the 

members of Collaborative Engineering Laboratory at the 

University of Tokushima. 

References 

[1] Ito, T. 2004, Vibration-based interface for remote 

object manipulation in video conference system, 13 th 

IEEE International Workshop on Robot and Human 

Interactive Communication, Kurashiki, Japan, pp.295- 

300. 

[2] Ito, T. 2004. (ed. K. Chen, D.Webb and R. March), An 

approach of virtual environment for planning of implant 

placement, Advances in e-Engineering and Digital 

Enterprise Technology. Pp.355-362. 

[1] Johanson, B., A. Fox, T Winograd. 2002. “The 

Interactive Workspaces Project: Experiences with 

Ubiquitous Computing Rooms.” IEEE Pervasive 

Computing Magazine April-June, 1/2:71-78. 

[2] Milne, A., T. Winograd. 2003. “The iLoft Project: A 

Technologically Advanced Collaborative Design 

Workspace as Research Instrument.” Proc. Int’l Conf. 

on Engineering Design ICDE2003, Stockholm, August, 

pp-no.1657. 

[3] Ponnekanti, S. R., B. Johanson, E. Kiciman and A. Fox. 

2003. “Portability, Extensibility and Robustness in 

iROS.” Proc. IEEE International Conference on 

Pervasive Computing and Communications 

(Percom2003), Dallas-Fort Worth, TX. March. 

[4] Telemedical.com home page URL 

http://www.telemedical.com/Telemedical/Products/vidc 

onf.htm

XML-based Product and Process Data Representation for Distributed Process Planning 

Dušan N. Šormaz, Jaikumar Arumugam, Narender Neerukonda 

Department of Industrial and Manufacturing Systems Engineering 

Ohio University, Athens, OH 45701 

Abstract 

The need to exchange product and process data between different software applications has been identified 

across manufacturing industries. An efficient model for communications between CAD, CAPP and CAM 

applications in distributed manufacturing planning environment has been seen as key ingredient for CIM. 

The paper proposes neutral, XML-based representation for product data required for process planning and 

for process data required for downstream manufacturing applications (like cell design, and scheduling). The 

representation is based on established standards for product data exchange and serves as a prototype 

implementation of these standards. The procedures for writing and parsing XML representations have been 

developed in object-oriented approach, in such a way that each object from object oriented model is 

responsible for storing its data into XML format. Similar approach is adopted for reading and parsing of the 

XML model. Parsing is preformed by a stack of XML handlers, each corresponding to a particular object in 

XML hierarchical model. This is very useful approach for direct distributed applications, in which data are 

passed in the form of XML streams to allow on-line communication. The components of the XML model for 

product and process data are explained in the paper, the mechanism and procedures for writing and parsing 

the XML model are described. The feasibility of the proposed model is verified in a sample scenario for 

distributed manufacturing planning that involves feature mapping from CAD file, process selection for 

several part designs and selection of alternative processes using scheduling criteria. 

Keywords: 

1 INTRODUCTION 

Computer Integrated Manufacturing (CIM) is still moving 

target in industrial and manufacturing engineering 

research and applications. There are numerous reports 

outlining individual islands of automation within the CIM 

model, with a significant effort in research papers being 

focused on a particular task, and much less effort devoted 

to integration of these manufacturing engineering and 

planning tasks. Integration involves the transfer of data 

between applications, but also should focus on data and 

model integrity, distributed processing of the data, 

incorporation of knowledge into the planning tasks, and so 

on. Computer aided process planning (CAPP) is rightly 

seen as an integration fabric for CIM with its relations to 

design (CAD), manufacturing (CAM), and scheduling 

tasks. By virtue of being an integrator between CAD, 

CAM, and scheduling, process planning involves decision 

making process on various levels of details with the final 

goal of generating feasible and/or optimal process plan. 

The need for integration requires that such model is 

transparent between these tasks, that it can be easily 

saved, transferred, and re-created as needs arise. 

This paper proposes such neutral data model in the form 

of XML model and describes its details and application in 

manufacturing process planning. The paper is organized 

as follows. Section 2 describes previous work in process 

plan representation and modeling. Section 3 describes 

process planning representation object model, which 

includes major entities and relations between them. 

Section 4 explains process planning XML schema model 

that is built for previous object model. Section 5 describes 

a flexible mechanism for writing data in XML format and 

parsing XML streams into object model. Section 6 

describes a sample scenario of integration between 

process selection and scheduling using data in XML 

format. The paper ends with conclusions and the list of 

references. 

12 

2 PREVIOUS WORK 

Data and knowledge representation in process planning 

have received significant interest in research. An early 

work on ALPS [1] proposed a graphical representation for 

manufacturing processes and means for specifying serial, 

parallel and concurrent tasks. Since then, several papers 

addressed knowledge representation, for example, using 

frames and rules [2] or an object-oriented data model [2]. 

International standard for product data exchange (STEP) 

has been developed to enable data transfer between 

applications that includes process data [8]. Recent results 

are in generation of the Process Specification Language 

(PSL) as a neutral format for the specification of process 

representation and exchange of different ontologies or 

semantics between various domains. Development of 

process planning specific, NC data within STEP standard 

has been described in [3]. Work on XML [6], as a very 

flexible language that transfers both data and their 

description (metadata) prompted its widespread use in 

many research efforts. Paper [12] discusses similarities 

and differences between STEP and XML and proposes 

their convergence. 

3 PROCESS PLANNING REPRESENTATION AND 

MODELING 

Manufacturing planning object model developed in this 

work is based on process planning object model proposed 

in [2], which described a data model for representation of 

process plans based on the different activities involved in 

manufacturing. A process planning representation model 

facilitates the development of algorithms for 

manufacturing problems like sequencing, scheduling etc. 

by reducing the over-all algorithm development time. The 

model accommodates a variety of data that may be 

needed in manufacturing planning algorithms. 

Components of the model are manufacturing planning 

model, feature object model, and process object model 

and they are described in this section.

3.1 Manufacturing planning object model 

Manufacturing planning object model is shown in Figure 1. 

The model is based on analysis of product and process 

design entities and includes hierarchical representation of 

manufacturing activities, that has manufacturing 

processes as its leaves, collection of manufacturing 

features and corresponding manufacturing processes, 

and a collection of machines used in the manufacturing 

system. 

A manufacturing activity represents the core of the model. 

Any activity that contributes to the manufacturing of a part 

is called as a manufacturing activity. A manufacturing 

activity has attributes like manufacturing cost, 

manufacturing time, member process etc. A 

manufacturing activity is subdivided as a part activity, a 

machine activity or a tool direction activity. A part activity 

describes the process plan for a part. There is an 

association relationship between part activity and part. 

Each part activity is associated with a part. A process plan 

for a part is a collection of the machines through which a 

part has to pass through to be completely manufactured. 

The part in turn can have multiple alternative process 

plans. The machining process of a part on each machine 

in its process plan is represented by a machine activity. 

Each machine activity is associated with a machine 

object. There is an aggregation relationship between part 

activity and machine activity. Each part activity is a 

collection of machine activities. The part object has 

attributes like a collection of its alternative process plans, 

part material, features list, process list etc. The machine 

object has attributes like machine name, number of units, 

usage frequency, SICGE code etc. Each machine activity 

is a collection of tool direction activities. A tool direction 

activity holds directional information about a machining 

process. The member process attribute of manufacturing 

activity is used to store the aggregations of a 

manufacturing activity. Thus, the member process 

attribute of a part activity holds a collection of machine 

activity. The member process attribute of a machine 

activity holds a collection of tool direction activities. The 

representation model for the cellular manufacturing shown 

in Figure 2 is built on the basis of the process plan 

Figure 1.Process plan representation model for manufacturing 

representation model and serves as its extension for cell 

formation. 

Figure 2. Cellular manufacturing representation model 

The core of the cellular manufacturing model is the 

manufacturing system. A manufacturing system is a place 

where parts are manufactured on machines. The main 

task of cell formation is the partition of the parts in the 

system into part families and machines into machine cells 

to reduce intercellular traffic. The objective is to find a 

partition for machines into machine cells and parts into 

part families in such a way as to minimize intercellular 

traffic. 

3.2 Feature object model 

Feature object model represents a hierarchical 

representation of various machining features with 

inheritance relations within it. The model is shown in 

Figure 3. The major class is MfgFeature that abstracts all 

common properties for all features. Properties at this level 

include feature name, containing part, tolerance data, list 

of alternative processes, and precedence relations. 

MfgFeature class is extended into several subclasses that 

correspond to machining feature types found in 

mechanical prismatic parts (such as Hole, Slot, and 

Pocket). These classes model properties of particular 

feature type and include different dimension parameters, 

and process capability data. However, model properties 

on general, feature ,level are of generic nature and can be 

applied by extending this model to other domains (like 

rotational parts, sheet metal parts, and so on). Model also 

contains several auxiliary classes that serve the purpose 

of generating various feature relations and displaying 

them to user. 

13

javax.swing.filechooser.FileFilt 

SaveFilter 

Exception 

UnsupportedFeatureExcept 

TableCellGenerator 

mfgFeatureInteractorGenerat 

TestPocket 

java.awt.event.ActionListen 

...actionAdapter 

Pocket 

Hole 

+HolePanel 

3.3 Machining process object model 

The knowledge about processes is also represented in an 

objected-oriented model. In order to map this knowledge 

representation, the machining processes are categorized 

based on their characteristics. Figure 4 presents this 

hierarchy in a UML based model. 

The class MfgProcess represents the most generic 

process class, i.e., a process with the common data 

shared by the rest of the processes. Further distinctions 

are carried out based on the process characteristics such 

as hole making processes and profile generating milling 

processes. The hole making processes are further divided 

into CoreMaking, HoleStarting and HoleImproving 

processes, while the Milling process has as sub-type 

java.awt.event.ActionListen 

...actionAdapter MfgPartMode 


MyMfgPartModel 

mfgFeatureNeighborGenera 

ImpObject 

MfgFeature 

MfgFeatureContentHandle 

ImpXmlHandler 

MfgFeatureSubclassHand 

Figure 3. Class diagram for features 


MfgFeaturePrecedenceGenera 

Slot 

+SlotPanel 

Figure 4. Machining process hierarchy 

14 

FeatureSet 

javax.swing.filechooser.FileFilt 

PartFilter 

InstanceFeature 

CheckBoxListene 

DovetailSlot TSlot 

LinearFeatureSe CircularFeatureSe 

EndMilling process. These generalized classes are 

implemented as abstract classes and are shown with 

italicized titles in Figure 4. The classes under these 

umbrella classes are for representing the actual 

machining processes (for example, TwistDrilling, 

EndMIllingSlotting, or FaceMilling). Therefore, based on 

the inheritance, EndMillingSlotting process acquires 

process information from ‘Endmilling’, which further leads 

to the parent class MfgProcess’. The following 

paragrpahs give a brief description about the MfgProcess 

and EndMilling classes., while description of other classes 

is part of working documentation. 

MfgProcess: This class contains member variables such 

as feature, stock, workpiece, cutting parameters,

constraints, tool, and tool path. These listed variables are 

used in every inherited process class as every process 

contains these components of machining process. Also, 

this class carries the GUI components for showing 

process information and a graphical interface to display a 

process. 

EndMilling: The implementation for visualization is mainly 

concentrated on end milling operations, so this class 

contains some graphical components. This class provides 

methods to prepare the scene graph, which carries nodes 

to display machining process components and the 

animation for visualization. The GUI required for eventbased 

interaction with the virtual world displayed is 

provided in this class. The subclasses --slotting and 

peripheral end milling-- use this implementation with 

distinctive modifications in tool approach for machining. 

4 PROCESS PLANNING XML SCHEMA MODEL 

This section discusses the persistent and neutral data 

storage mechanism for the process plan representation 

model (PPRM) based on XML. It describes the XML 

schema which is the basis or template for process plan 

data generation, and provides few examples of XML data. 

A schema is a set of rules that have to be satisfied in valid 

XML document [6]. Figure 5 shows the XML Schema 

definition for the process plan representation model as 

shown in a tool for XML development XMLSpySuite. 

Figure 5.XML schema for PPRM from XML Spy Suite. 

The root element of the schema is that 

holds all information about a factory model. This element 

holds two collection elements and 

. The element is a collection of 

part models represented by the element. 

The is a collection of the available 

machines represented by element. Each 

element has a list element called 

, which is a collection of alternative 

process plans, for the enclosing part element. These 

alternative process plans are represented as 

elements on the part model. Each 

element has a 

element. The element is a 

collection of all the elements in the 

enclosing process plan. Each element 

encloses a element that holds information 

about the machine on which the machining operation is 

done. Each element holds a 

element, This element holds 

the spatial information about the tools used in this 

machining activity. in turn, holds 

element as reference to a machining 

process. The hierarchical relationship between parts, 

machines, part activities, machine activities, tool activities, 

features and processes is in such way incorporated into 

the XML document. 

Information on features and processes may be included in 

few levels of detail. For feature modeling, geometric 

reasoning and process selection all details are needed, so 

element actually carry the name of the feature or process 

class, with inclusion of all details. Example of this model 

for two features is shown in Figure 6. Feature tags are 

actually class names, and feature have all details of their 

dimensions stored. 

Figure 6. Detailed XML data for features 

Different XML elements are used when feature and 

process data are used in cell formation or scheduling, as 

will be shown in detailed example in section 6. 

5 MECHANISM FOR WRITING AND PARSING XML 

DATA 

This section describes flexible mechanism for generation 

of XML data from Java and reconstruction of class 

instances from XML code. 

5.1 XML data generation mechanism 

An object-oriented data generation framework was 

constructed for the purpose of generating XML data for 

factory models based on the process plan representation 

model. The different entities whose XML data needs to be 

generated for the factory model include MfgPartModel, 

Machine, PartActivity, MachineActivity and 

ToolDirectionActivity, MfgFeatue and MfgProcess. All the 

Java classes built for the process plan representation 

model are capable of generating self-describing XML 

data. We propose two alternative ways for achieving this 

goal: a) each class implements methods for XML data 

generation, or b) XML data generation is delegated to 

special XML writer classes. 

The first approach was used for Machine class as shown 

in Figure 7 where the code for its method writeXML() is 

given. The Machine class holds data about the machine 

like name, number of units, usage frequency, SICGE code 

etc. It is this state data for each machine that needs to be 

saved into XML. 

15 

 

 

 

 

 

 

Figure 7. Machine Java class showing XML data 

generation mechanism

To establish uniformity, a method called writeXML() with 

the following syntax is defined fro each class using this 

way: 

public void writeXML (StringBuffer xmlTarget, 

String indent) 

This method is called on the Java class whenever data 

about the state of an instance of the class has to be 

generated in the form of XML. The argument xmlTarget 

holds all the factory model information generated in XML 

form until the Machine class method was called. The 

generated XML data from the Machine class is then 

included in the xmlTag variable. The indent argument is to 

generate properly indented XML code to improve 

readability. 

Alternative approach is to implement XML writer class for 

each PPRM class. The XML writers write out XML data at 

a given state of manufacturing process planning .The 

PartWriter writes part material and part activities. It calls 

FeatureWriter to write the feature data. 

The feature writers also work in the same manner. The 

FeatureWriter writes the dimensional details like radius, 

depth, axis and axis-point about the feature. It gives a call 

to the PropertyTableWriter which writes the tolerance 

details like surface finish, roundness, true position and 

straightness. It also calls ProcessWriter that writes data 

fro alternative processes like, class name (type), process 

name, cutting parameters, machine name, process time 

and cost. This chain of XML data writing is shown in 

Figure 8. 

Part Writer 

Feature Writer 

Process Writer 

Part name 

Part material 

Batch size 

Call feature writer 

Call activity writer 

Feature name 

Radius 

Depth 

Axis 

Axis point 

Call tolerance writer 

Call process writer 

Tool Material 

Depth of cut 

Cutting parameters 

Machine name 

Process time 

Process cost 

Figure 8:XMLWriters for processes 

It should be noted that during the writing process writers 

verify the state of the process planning and write only data 

that is available. For example, at the end of feature 

mapping only feature details are written, and the writer 

does not write anything about processes or tolerances. 

5.2 XML data parsing 

XML data generated from PPRM holds all the information 

necessary to reconstruct the Java class instance, 

including data about all the parts in the factory model, the 

machines in the factory model, the routing information for 

all the parts etc. XML parsers can be used to reconstruct 

instances of Java classes from the XML data. There are 

two kinds of parsers available for XML data parsing [6]: 

SAX parser (Simple API for XML parsing) and DOM parse 

(Document object model). DOM views an XML document 

16 

as a tree structure and loads the entire document into 

memory. It builds parent-child relationship between 

nested elements. The DOM API provides standard 

methods for querying the XML document and 

reconstructing Java objects. SAX is an event-based 

parser that fires different events based on the element 

parsed. A SAX parser, unlike the DOM parser, does not 

maintain a default model for the parsed data. If a listener 

can be set on the SAX parser to listen to parsing of 

specific tags, then whenever the corresponding element is 

parsed, the listener can construct a Java instance of the 

class corresponding to the element. The SAX parser used 

in Factory Model Generator is an instance of the Java 

class (javax.xml.parsers.SAXParser). This parser is used 

to construct factory models from the saved XML files. The 

listener for this parser is called as content handler and is 

an instance of the Java class 

(org.xml.sax.helpers.DefaultHandler). Every object (or 

class) in the PPRM is responsible for defining and 

implementing a content handler that can reconstruct 

instances of the Java class from the XML element parsed. 

For example, for the Machine class, the content handler is 

defined as an inner class and an instance of the content 

handler can be obtained by calling a method called 

getSAXHandler(). XML data can then be parsed with this 

content handler. 

The content handler provides three methods that are used 

to reconstruct the wrapper class instance: startElement(), 

endElement(), and characters(). The content handler 

class for parsing XML data of the machine XML data is 

shown in Figure 9. 

Figure 9. Parsing machine XML data with a content 

handler 

An instance of the SAXHandler class is capable of 

generating events when a element is parsed 

in the XML document. Separate events are generated 

corresponding to the parsing of , 

and character data. The startElement () 

method is called when the SAXParser parses the 

tag. Since all the information in the 

element is available as attributes of the element, the 

values corresponding to the attributes can be obtained 

from the atts argument of the method, as shown in the 

code above. Thus, the Java object mac corresponding to 

the element tag parsed is constructed. The 

characters method is fired when characters are parsed 

between the and the . 

Since the tag shown above does 

not have any characters, this method never gets fired. 

The endElement () method gets fired when the

tag is parsed by the SAXParser. This method 

can be used to deal with the instance of the Machine 

class constructed in the startElement () method. For 

example, this instance can be added to the machineList 

variable of the MfgSystem object, which holds this 

machine. The content handler of the MfgSystem class has 

already constructed the MfgSystem object by the time the 

Machine element is parsed. This MfgSystem object is 

passed to the machine content handler as a variable 

called parent. 

5.3 Hierarchy of XML Handlers 

In order to accommodate method described in the 

previous section, it was necessary to provide a 

mechanism for parsing a single XML file (or stream). This 

is responsibility of ImpXmlReader class. This class 

accepts the XML stream and initial handler, and delegates 

XML parsing task to this handler. As different XML 

elements require different handlers, ImpXmlReader keeps 

a stack of handlers, so that when new handler is invoked, 

the old is kept stored in the stack. When parsing of one 

PPRM object is completed, the last handler is recalled 

from the stack and it continues parsing. At the end, 

parsing of the file is completed by the very first handler 

that started it. Thus, the entire PPRM model instance is 

constructed from the XML data. 

In order to better explain the above described process, 

and example of parsing a single part file is shown in 

Figure 10. The file is given to ImpXmlReader, which starts 

parsing using the PartHandler. The PartHandler starts 

interpreting tags (or elements) in the XML file. As soon as 

it encounters the tag it switches the 

handler and sets it to the FeatureHandler class which 

continues parsing the file further. The PartHandler is 

pushed into the stack of ImpXmlReader. The feature 

handler instantiates the feature object and populates it 

with data using setter methods as explained earlier. The 

feature handler as it parses and reads the 

tag shifts and sets the handler to 

PropertyTableHandler class. This class sets all the 

tolerances on the current feature. On parsing the closing 

tolerance tag the handler is switched 

back to the FeatureHandler. 

XML tag Handler Action 

PartHandler 

Switch to FeatureHandler 

 

 

 

 

Switch to 

PropertyTablehandler 

Switch back to 

Featurehandler 

Keep the same handler 

Switch back to Part handler 

MfgPart Object 

instantiated 

Feature object instantiated 

feature dimesions set: 

axis, axis point, radius, 

depth 

Tolerances to Feature 

set: truePosition, 

roundness, positive 

tolerance, sets surface 

finish, etc. 

Feature added to part 

MfgPart Object completed 

Figure 10.Functioning of XML Handlers. 

The feature also is added to MfgPartModel object. Finally 

when all features are done and the tag is 

encountered the handler is reset to PartHandler, which 

completes the parsing the file. 

This approach provides for great flexibility in parsing XML 

data. For example, in cell formation, layout design, 

capacity planning, or scheduling applications, all data may 

be stored in a single file that corresponds to the whole 

factory model. In that case, MfgSystemHandler starts 

parsing, and it calls PartHandler and MachineHandler at a 

corresponding tag in the file and resumes parsing when 

they finish the work. This approach also provides fro a 

scenario in which parts may be parsed from different files 

or other sources, and then combined into single file, or 

parsed in a single file and then split for individual 

processing in some application. 

The existence of separate XML writers and handlers 

provides for another extensibility of the approach. The 

XML writer and handler can be specified at run time, so 

that it is possible to parse data in one format and then 

store it back in a different format. Utilization of this method 

is for applications where not all details about the model 

are required, but only some subset. Also in cases where 

applications already exist with particular structure, this 

structure may be used to extend the applicability. 

We have tested this approach in a rule-based process 

selection system in which communication between 

IMPlanner [12] (the prototype in which this approach has 

been implemented and tested) successfully implemented 

a different set of writers and handlers, with the need to reimplement 

any other class and we were able to utilize 

system in this case. 

6 INTEGRATION OF PROCESS PLANNING AND 

SCHEDULING – AN EXAMPLE SCENARIO 

Utilization of XML-based process plan data model will be 

demonstrated in distributed integration of process 

planning and scheduling. Scenario is the following. For a 

group of designed parts it is necessary to perform process 

planning and select the process plan from several 

alternatives in order to achieve shortest processing time 

(makespan) on a group of available machines. Cad 

models are available for these parts. 

This scenario will be illustrated on several mechanical 

parts from NIST design repository [5]. Four different parts 

were chosen as shown in Figure 11. For each part CAD 

model was modified to feature based model to 

accommodate our approach to CAD/CAPP integration. 

The procedure will be illustrated on AES94 model. 

Feature based model was created in Unigraphics CAD 

package (Figure 12). Each model has been separately run 

by feature mapping application (details are in [9]) as 

shown in Figure 13. The result for all models is saved into 

XML file with the full feature details (Figure 14). Process 

selection is done by rule-based system [10] which was 

developed for an industrial partner, with results shown in 

Figure 15. For scheduling application, XML files for all 

models are used in simplified version (scheduling is not 

concerned with feature details, but only with process 

times and machines), so different writers and parsers are 

used (Figure 16 shows one input file). Scheduling 

application used LP approach for selection of 

simultaneous selection of alternatives and their 

scheduling [11], its results are shown interactively (Figure 

17) and saved into XML file, which is shown in Figure 18, 

successfully completing the whole scenario. 

7 CONCLUSIONS 

This paper has demonstrated the flexible use of XML data 

representation in process planning modeling and 

integration. The XML schema has been developed for 

validation of XML process plan files. 

17

Name AES94 Anc101 

Model 

Number of 

features 

12 20 

Name Switcharm Scfdemo04 

Model 

Number of 

features 

Figure 12. Feature based model in CAD system 

Flexible approach for writing XML data from PPRM and 

parsing data back has been explained. Benefits of this 

approach were demonstrated in few small examples. 

Scenario for the use of proposed XML based model in 

integration of feature mapping, process planning, and 

scheduling application is demonstrated on simple 

example. 

9 18 

Figure 11. Models used in the example scenario 

18 

Figure 13. Feature mapping application 

8 REFERENCES 

[1] B. A. Catron, and S. R. Ray, ALPS: A Language for 

Process Specification. International Journal of 

Computer Integrated Manufacturing, 4, 105-113. 

1991. 

[2] J. Y. Park, B. Khoshnevis. A real time computer 

aided process planning system as a support tool for 

economic product design. Journal of Manufacturing 

Systems, 12(6):181-193, 1993.

Figure 14 XML file as a results of feature mapping 

Figure 15. XML file as a result of process selection 

Figure 16. XML file as input to scheduling 

Figure 17.Scheduling application with the solution 

19 

Figure 18. XML as a solution to scheduling algorithm 

[3] D. N. Sormaz and B. Khoshnevis, Process Planning 

Knowledge Representation using an Object-oriented 

Data Model, Int. Journal of Computer Integrated 

Manufacturing, Vol. 10, No. 1-4, p. 92-104, 1997. 

[4] C. Schlenoff, A. Knutilla, S. Ray, Proceedings of the 

Process Specification Language (PSL) Roundtable, 

National Institute of Standards and Technology, 

Gaithersburg, MD, 1997 

[5] Regli, William C, Gaines, Daniel M., A repository for 

design, process planning and assembly, Computer 

Aided Design, Vol. 29, No.12 p. 895-905., 1997. 

available on-lone at the URL: 

[6] 

http://edge.mcs.drexel.edu/repository/frameset.html 

S. Holzner, Inside XML, New Riders, Publishing, 

Indianapolis, Indiana,2001. 

[7] S T Newman, Integrated CAD/CAPP/CAM/CNC 

Manufacture for the 21st Century, Invited paper 

Flexible Automation and Intelligent Manufacturing, 

FAIM2004, Toronto, CA, 2004. 

[8] STEPTools web site www.stepttols.com, September 

2004. 

[9] J. Arumugam, IMSE Department, Analysis of 

Feature Interactions and Generation of Precedence 

Network for Automated Process Planning, MSc 

thesis, Ohio University, 2004. 

[10] A. Wadatkar, Process Selection fro Hole Operations 

Using a Rule Based Approach, Msc thesis, Ohio 

University, 2004. 

[11] R. S. Harihara, Modeling Scheduling Algorithms with 

Alternative Process Plans in OPL, Msc thesis, Ohio 

University, 2004. 

[12] D. N. Sormaz, J. Arumugam, S. Rajaraman, 

Integrative Process Plan Model and Representation 

for Intelligent Distributed Manufacturing Planning, 

International Journal of Production Research, Vol. 

42, No. 17, p. 3397 - 3417, 2004. 

[13] J. Lubell, R. S. Peak, V. Srinivasan, S. C. Waterbury, 

STEP, XML, and UML: Complementary 

Technologies, Proceedings of DETC’04, ASME 2004 

Design Engineering Technical Conferences and 

Computers and Information in Engineering 

Conference, Sep.28-Oct. 2, 2004, Salt Lake City, UT

PART 2 

INTEGRATION & PRODUCT 

VARIANTS AND COMPLEXITY 

20

The integration of CAE into the 

Complete Manufacturing Workflow 

Integration of ECAE into workflow process Tony Ward / EPLAN Sales / 15 November 2004 

21 

1

Acceptance of CAE in the market 

• The need for E-CAE (Electrical Computer Aided 

Engineering) is now far more accepted in the market 

than 10 years ago 

• At that time, the mechanical drawing office would 

often select the CAD system that a company would 

use, the electrical drawing office having to make best 

use of a 2D CAD solution 


22

Problems with 2D CAD design 

• If using a 2D CAD system, the process of electrical 

design is far less automated. 

• Time spent on producing electrical documentation is 

a main part of the project. 

– Circuit diagram 

– Cross-References 

– Possible terminal diagrams 

– Possible bill of materials 

• CAD = Computer Aided Drawing, all these 

evaluations required for electrical engineering must 

still be produced manually 


23

Compared to conventional 2D CAD 

E_CAE solutions can deliver 

Reduction in errors and project design costs through: 

•Rapid Generation and modification of schematics 

•Automatic generation of project documentation 

such as terminal, wire, cable lists and BoM’s 

•Easy modification process greatly reducing time 

and errors 

•Single data entry solution reducing errors 


24

Problems with 2D CAD design 

CAD - Alone is not the key for success 

• Fact: Computer based drawing does not bring all 

the benefits alone. 

Cost & Time intensive evaluations missing. 

Evaluations still generated manually 

• Higher demands on documentation because of: 

– Automisation of machinery /plant. 

– Greater demands from quality standards, 

customer expects error free documentation 

– Optimised document flow 


25

There is a better way! 

• Database driven E-CAE systems are now much more 

widely used, particularly in continental Europe but are 

still often seen as ‘a design island’ within the overall 

project management process. 

• This leads to multiple entries of the same data in 

different systems within the process 

• A far more efficient approach is to optimise this 

process through single data entry and interfaces 


26

PLC I/O 

The typical disjointed approach 

might consist of a workflow like this: 

Bus info 

= manual process 

PDM 

CAE design 

MRP 


27 

Drawings 

Panel Build 

Customer 

Maintenance

What is being transferred and how? 


A verbal transfer of 

information such as 

PDM 

project number, 

completion date, 

customer details, 

CAE design project value, 

machine details. 

Name of CAE project, 

date drawings 

completed, revision 

information etc. 


28

PLC I/O 

Bus info 



PDM 

CAE design 


29 

An I/O list may be 

printed as a Word 

or ASCII file and 

passed to the 

designer who then 

duplicates this 

information in the 

schematic. 

Sometimes the 

process works in 

reverse


A parts list is created 

by the designer then 

printed out to be 

entered into a stock 

control / purchasing 

system. A separate 

bill of material may 

be printed for panel 

build, customer and 

maintenance 

requirements 

CAE design Drawings 

MRP 



30 

Panel Build 

Customer 

Maintenance


Paper drawings are 

printed out for the 

panel builder to work 

to, DXF files may be 

created for the 

customer and 

maintenance, losing 

all the intelligence 

built into the E-CAE 

design. 


CAE design Drawings 


31 

Panel Build 

Customer 

Maintenance

PLC I/O 

Bus info 

Now make a modification, can you 

be sure the change is implemented 

in all areas? 

PDM 

CAE design 

MRP 


32 

Drawings 

Panel Build 

Customer 

Maintenance

There is a better way! 

• By creating interfaces between the different solutions 

and transferring data by means of spreadsheets or 

the lowest common denominator – an ASCII file! 

• This ensures a single point of entry for all data, either 

in the PDM system, PLC programme or the electrical 

design. 

• Greatly improves the workflow process and reduces 

time and, more importantly, errors in the manual 

transfer of data 


33

The new concept, workflow process! 

PLC 

PDM CAE 

Bus 


34 

Panel 

Language 

MRP 

Maint. 

Customer

To summarise: 

E-CAE solutions already offer 

• Significantly improved performance through time 

saving and quality schematics 

• Reduction in errors and reduction in overall project 

design costs in excess of 75% 

• Quality support documentation (auto generated). 

• Improved quality of design and documentation 

• Now, take it one step further and integrate into the 

workflow process for even greater efficiency 

• Let the DESIGNER DESIGN! 


35

Aachen Technical University, Laboratory for Machine Tools and Production Engineering 

Release-Engineering: 

An Innovative Approach to Handle Complexity of Mechatronic Products 

Introduction 

G. Schuh, J.C. Desoi, M. Lenders, V. Witte 

Abstract: 

Complexity Management has emerged to an essential part of management in the automotive industry. Only if all drivers 

of complexity are well balanced, an economic optimum of variety can be achieved. Current automotive trends increasingly 

accelerate the deterioration of mechatronics products especially when compared to more recent rival products. 

Driven by the advent of innovative vehicle applications, contemporary automotive mechatronic architecture has reached 

a level of complexity which requires a technological breakthrough in order to manage it satisfactorily and to fulfil the 

customer and legal requirements. The inevitable complexity drivers in dynamic automotive markets will remain the increasing 

customer orientation, rising model diversity and the increasing software and mechatronics shares. The presented 

Release-Engineering approach allows to control and to manage innovation programmes by rather suitable, dedicated 

R&D efforts on emerging variants over the product lifecycle. 

Keywords: Product Programme Management, Development, Mechatronics 

The complexity of R&D projects in the automotive industry 

is still increasing. The rising number of mechatronics 

applications and the growing number of derivative product 

variants stress the factor complexity even more. In fact, in 

81% of products the number of variants was found to 

increase over the product life cycle in a recent WZL-study 

(with an increase of variants by the factor 2 or more in 

43% of the cases). At the same time, only 19% of the 

companies perform a systematic integration of part commonality 

schemes into the product development process 

with a high priority [1]. However, companies within a 

premium market are required to offer a greater product 

variety to maintain market share. The challenge to handle 

the increasing number of models and derivative products 

to penetrate market niches becomes even more complex. 

Thus, the balancing act between differentiation due to 

customized product offerings being launched at ever 

shorter intervals and concurrent amortization of R&D 

efforts becomes one of the main challenges. 

Opportunities and needs in automotive innovation 

Current trends increasingly accelerate the deterioration of 

mechatronics products especially when compared to more 

recent rival products. Driven by the advent of innovative 

vehicle applications, contemporary automotive mechatronic 

architecture has reached a level of complexity 

which requires a technological breakthrough in order to 

manage it satisfactorily and to fulfil customer requirements. 

A number of studies have focused on the speed and 

the productivity of individual projects in automotive development 

[2,3,4,5,6,7]. A common finding is that in order 

to shorten development lead times a relatively projectoriented 

organization with cross-functional coordination is 

essential. Strong project managers facilitate quick completion 

of a project by integrating different functions within 

the project. This approach has led to successful individual 

projects but is not necessarily the most efficient one for 

considering linkages among multiple projects of an entire 

company. Companies understand the need to cope with 

today’s dynamics of change and to reduce complexityrelated 

R&D efforts while utilizing the potentials of elec- 

36 

tronics and embedded software. However, the process to 

achieve this is usually not clear. Especially mechatronics 

and software driven R&D managers have not fully discovered 

that uncoordinated innovation and product programme 

management itself can produce unnecessary 

complexity costs. They realize that they did not examine 

thoroughly enough what kind of active standards in terms 

of basic design architectures they have to provide to 

achieve a cost-effective development strategy over the 

programme lifecycle. The obvious consequence is that the 

proactive management of product variety and complexity 

generation during the product life cycle and the associated 

R&D efforts is a key factor to economic success. 

The mechatronisation of components, product proliferation 

and decreasing model cycles are essential complexity 

factors in product development. All factors are challenges 

especially in European premium product and component 

development. As a consequence, the expected economies 

of scale associated with the launch of new variants are 

often overestimated and resulting dis-economies of scope 

are underrated [8]. 

Origins of Release-Engineering 

The Release-Management approach is derived from Software-Engineering 

and can be carried forward to automotive 

(mechatronical) systems and significantly reduce 

unnecessary product complexity. Two major reasons 

stimulate software developers to coordinate software 

development in releases. They differ in their temporal 

domain – proactive complexity reduction over the lifecycle 

as well as R&D efficiency enhancement over the 

different developing stages. Without a release-based product 

programme planning, either software would always be 

obsolete or none of the sophisticated hardware components 

could be installed. Product releases are also built so 

that they are approved for various platforms and markets 

[9]. Thus, one release represents a harmonised bundle of 

enhancements to a product’s functionality. The cycles in 

which releases take place are planned ahead; i.e. during 

the Windows 2000 development, Microsoft enforced a 

strict schedule for submitting revisions. Here, a release 

typically equals up to 250 changes [10].

Engineering in releases could feature a sustainable way to 

control and leverage emerging complexity. Especially 

those components for which cyclicity of technological 

evolutions comes into the fore – e.g. telemetrics, engineelectronics, 

active chassis frames – are predestined for a 

Release-Engineering approach. These components and 

their interfaces can benefit from an early coordination of 

following adoptions and innovations in derivatives or 

contiguous product families. The best possible functionality 

could be combined with the best possible robustness of 

the entire system. 

Perspectives of Release-Engineering 

Release-Engineering features advantages on three different 

levels of product development. It substantially improves 

product programme planning, requirement specifications 

management and change management efforts during advanced 

development stages. 

Utilizing Release-Engineering for product programme 

planning means evaluating the market acceptance of new 

innovations beforehand and thereby actively planning the 

company’s innovation roadmap. The impact of Release- 

Engineering in this context means to focus new innovations 

over the whole product programme on one’s unique 

selling proposition. By a well-targeted timing of the introduction 

of innovations, innovations are assigned to certain 

Release-intervals and the perceived level of innovation can 

be controlled. 

Due to the parallel existence of increasingly divergent 

component life cycles within today’s automotive systems, 

product architecture specifications need to be kept flexible 

for decisive functions at the right point of time. The decisive 

factor is the successful identification of the underlying, 

“ideal” component clock-speeds. 

The aspect of consolidating changes represents a central 

aspect of the methodology of Release-Engineering. 

Changes are to be executed only within the scope of predefined 

release-intervals. The synergy effect of the consolidation 

of changes consists of the savings on reactive 

changes. Reactive changes potentially appear whenever a 

component is changed that is “coupled” to other components. 

The approach presented in this paper is focussed on the 

product release planning. The Release-Engineering approach 

for product management in general is based on the 

following two hypotheses. 

1. Hypothesis: Active regulation of the commonality 

degree within the product programme 

Only by actively integrating variant planning into the 

product development process it will be possible to generate 

a variant- and innovation triggered economic growth 

next to lucrative cost-structures. In case OEMs and suppliers 

are not able to consider future variants already within 

the concept phase, rising complexity cost will ruin the 

profitable efficiency [11]. High degrees of commonality 

and notable product innovations are not contradictory. The 

duty of conventional program planning is to aim at a 

sensible compromise between a mere increase of the newproduct-rate 

and the introduction of new technologies as 

opposed to the extensive reuse of existing technologies 

37 

and applications („[...] the rapid reuse among multiple 

projects of new technology may actually improve the 

overall newness or technological sophistication of a firm’s 

product offerings. Therefore, we can hypothesize that the 

negative impact on a firm’s market competitiveness 

should depend to some extent on the average design age of 

new products introduced into the marketplace” [3]). The 

success factor of covering the competitive ability of OEMs 

and suppliers is a consequent integration of variant planning 

into the development process, defined for the complete 

product programme lifecycle [12]. Successful control 

of product variety starts at the beginning of the development 

process and ends with the withdrawal of the product 

from the market. It requires tools and methods to control 

the evolution of product and part variants and the recognition 

and elimination of unnecessary variants from both the 

technical and economical point of view. Professional 

complexity management will be assisted by engineering in 

releases. 

2. Hypothesis: Active planning of intended innovations 

Innovation planning has to be focussed on long term 

ranges and also has to be implemented uncoupled of market 

cyclicity. As a consequence of shifting value-added 

share to first tier suppliers, the system and component 

creation will be one of the main triggers for innovation 

planning. The classical product development process will 

not be obsolete, but the creation of innovation grades will 

be defined by innovations on component level. Amortisation 

of R&D expenditure, calculated for component usage 

within different product lines, will be one of the valuable 

decision points for new innovation cycles. 

Focusing Automotive Development on Releases 

Automotive OEMs have recognized the complexity problem 

in the early nineties. Complexity increases R&Dcosts 

exponentially through exotic status of most of the 

resolving variants and reduces OEMs and supplier profits 

[13]. Existing cost calculation systems often do not 

support an identification of costs and R&D efforts according 

to the input involved. Especially the phenomenon of 

unconscious cross-subsiding of niche variants is problematic. 

Management of R&D activities by means of Release- 

Engineering means to glance at synergy potential according 

to upcoming product innovations. However, for trendsetting 

product innovations, it can be strategically sensible 

to cross-subsidize new solutions, even if they are overengineered 

for the actual project. 

To address the general complexity problem, companies 

have two possibilities: they can either make their processes 

and systems more efficient in order to handle increased 

variant variety throughout the whole company or 

they can concentrate on developing a product release 

strategy that is optimized to meet the market requirements 

thereby avoiding unnecessary variants and R&D efforts. 

The first approach seems currently to be most common 

among OEMs and suppliers, but is like curing the symptoms 

instead of killing the virus itself. It is really promising 

to keep focus on single project complexity management 

because it allows for keeping the internal product 

systems as simple as possible, thereby reducing internal 

error rates and allowing for more streamlined processes.

Furthermore, it leads to high shares of commonality on 

part usage rates, but impedes to find the really big treasures. 

The second possibility is focussing on the product programme 

planning with new innovation triggers, and seems 

to be the most probable way – according to the trend of 

rising added value on supplier’s side. The key success 

factor of complexity management can be found in the 

product planning phase. To control and manage the diversity 

of variants means to describe profits as well as R&D 

efforts not only for the basic project, but rather for the 

following application projects over the product lifecycle. 

The careful analysis of all relevant product programme 

lifecycle requirements as well as the definition of the 

product variants to be offered on the market is the enabler 

for cost-effective product programme development. Only 

companies that master aspects of product specific R&D 

efforts and realise the matching of following application 

project requirements will be on their way out of the dilemma 

of economies-of-scale in engineering departments. 

Requirements have not to be defined only for the basic 

project but rather have to be managed for all possible 

application projects in the future. Release-Engineering as a 

methodical principle means balancing the complexity 

drivers correctly over the product programme lifecycle on 

Multi-Project Management. 

The automotive development process is already well 

defined by gateways and milestones – but merely in perspective 

of a single project and only on a generic level. 

The creation of product variety and the timing of change 

programs over the product lifecycle are usually not implemented 

within these schemes. A stronger focus on 

change management within the context of release requirements 

will be a solution to proactively shape the future 

product variance formation (see Figure 1). Not every change 

cycle is absolutely necessary– an optimum ratio of interdependencies 

that have to be overcome and innovation 

contribution is important. 

Quantity 

Release 3 

Release 2 

Release 1 

Today gestern Yesterday 

Frequency 

scale 

Exotics Standard 

Exotics 

Figure 1: Dynamic generation of product variants 

Time 

Many product adoptions and changes obviously have to be 

inserted to cover development faults during the development 

phase or to shape product innovativeness over the 

product lifecycle. Due to distributed development responsibilities 

towards suppliers as well as towards wider product 

ranges this isolated approach is no longer feasible. 

Impacts of necessary and unnecessary changes as well as 

reactive changes have to be implicated in the long term. 

Automotive innovation frequency – as another reason for 

38 

necessary changes – is determined by a mix of innovation 

rates e.g. for engine, body & mechatronics/ electronics 

[14]. But innovation cycles are mainly different – 

electronic products rarely last longer than 18 months with 

performance doubling from generation to generation [15] 

where power train cycles could exist several years. Decoupling 

in releases is the solution to manage the dichotomy 

between rising R&D effort for product or component 

changes and necessary economies of scale of the entire 

product [16]. 

Reduction of R&D-complexity through 

Release-Engineering 

The methodology of Release-Engineering can be utilised 

to consequently structure the product program towards an 

“optimum” fulfilment of customer requirements. The preplanning 

of releases on behalf of configuration management 

allows a steady realignment of variants demanded by 

the markets without causing a disproportionate increase in 

variance. Within the framework of Release-Engineering, 

the rather ample “Major Releases” provide useful points in 

time to revise the significance of actual, current product 

variants and replace less relevant ones (see Figure 2). Configuration 

management plans releases so that the mandatory 

adjustment to changed customer requirements takes 

place. At the same time, the commonly observed disproportionate 

increase in product variety can be avoided. 

X 

A 

B 

Release Intervals 

Product 

Variance 

„Major Release“ „Major Release“ 

Figure 2: Continuous adjustment of product variance 

time 

R&D-efforts will be lowered by the time-wise bundling of 

component changes without affecting the diversity necessary 

marketwise yet including price sensitive configuration 

opportunities. More significant innovation programs 

by more dedicated R&D efforts on system variants could 

be achieved by governed variant reduction cycles. Disadvantageous 

features and according variants can be eliminated 

or replaced at each major release cycle. 

The Spiral Model of Software Development and its 

Implications for Release-Engineering 

The spiral model for software development is a model for 

cyclic, risk oriented development (see Figure 3). In general, 

there are no harmonised hard- and software-cycles in the 

computer industry, neither by generating nor by voiding of 

innovations. An active Release-Engineering thereby supports 

a continuous improvement of innovativeness over 

the software lifecycle. In the computer industry, software 

always has to be adapted to new hardware specifications 

or devices. Without a release-bases management, software 

developers could not react on new hardware technology or

and update the software. The principle development methodology 

can be carried forward to automotive mechatronic 

development, but here the major trigger for new product 

adoption are product characteristics or requirements defined 

for application or variant projects e.g. when downscaling 

product component innovations to lower product 

segments. Simultaneously, fix synchronisation points have 

to be defined and transferred to suited parts of the spiral 

process. 

Product development 

Example: Operating System 

Market 

Launch 

Release 1 Release 2 

Definition of 

synchronisation points 

Market Release 1‘ 

Launch 

Development of application 

Example: Office package 

Synchronisation point product development 

Synchronisation point development of application 

Identify targets, alternatives 

and restrictions 

Planning of 

next release 

Plan next cycle 

Release 1 

Assess alternatives, 

identify and minimize risks 

System specification 

Market Launch 

Concept for 

architecture 

Realise and verify 

chosen alternatives 

Figure 3: Cyclic synchronisation points within the product programme 

For example if actively regulated chassis components, 

originally developed for limousines, have to be adapted for 

other derivatives like station wagons or sports cars on the 

same platform, the dimensioning for maximum loading 

and deceleration values has to be adjusted. 

In industrial practice, it is very common to first develop 

the basic product instead of generating a whole building 

set. After start of production, the basic product is gradually 

supplemented by additional product features. After 

enabling the basic type, those applications cause an overproportional 

growth of development expenses. Objective 

of the Release-Engineering methodology is to consider the 

to-be product features within the basic product structure 

during early phases. The managing complexity methodology 

“Variant Mode and Effect Analysis (VMEA)” is 

applied to analyze the complexity origin and provides a 

structured and transparent survey of how variety is emerging 

[17]. Availability of information is one of the key 

requirements for the active process of managing releases. 

The procedure for release planning can be deducted from 

the VMEA methodology. The major step forward by 

means of Release-Engineering is to consider the dynamic 

evolution of variants over the product lifecycle. 

The VMEA uses two interpretations – the functional and 

the structural description of product variants. As can be 

seen in Figure 4, the functional structure of each variant is 

described by related product functions. Thereby it is important 

to define necessary product functions and their 

respective attributes, but not to focus too much about the 

later technical implementation. The features can be arranged 

according to their importance or other criteria, 

forming a so-called Feature Tree as shown in Figure 4. 

The right end of the Feature Tree represents all product 

types to be offered on the market and should be completed 

by sales volumes which then allows for the calculation of 

attribute usage analysis on all levels and branches of the 

Feature Tree. After finalization, this type list represents 

the diversity target for the product programme to be developed. 

39 

On the other hand, from a complexity management point 

of view the main question to be answered in this phase is 

how to technically implement the required product diversity 

most effectively and efficiently. The development of 

parts and modules is supported by the Variant Tree methodology 

(Figure 4). While similar to the Feature Tree, this 

methodology is used to control the internal complexity and 

to ensure that products are realized with the least variety 

possible on the parts and module level. The Variant Tree 

shows the development of diversity along the assembly 

sequence of any part or module. Parts information of 

existing parts (e.g. carry-over parts) can be loaded and 

new parts can be entered easily with all relevant information. 

In terms of Release-Engineering, the most important question 

is how the mutation of both structures over the product 

life cycle can be compromised. 

Product Structure 

E 

B B B 

B T T T B 

Functional Structure 

B471 

T T T T T T T 

10kW 

20kW 

Product Characteristics 

Keilr. 

Zahnr. 

Wasser 

Öl 

Wasser 

Öl 

3.0 

3.5 

3.0 

3.5 

3.0 

3.5 

4.0 

4.5 

3.0 

3.5 

4.0 

4.5 

100C 

100C 

120C 

100C 

120C 

100C 

120C 

120C 

120C 

120C 

120C 

120C 

120C 

120C 

System 

specification 

Concept for 

architecture 

Market Launch 

Release 1 

Variants 

Figure 4: Interdependencies of product functions and product structure 

Complexity management by means of Release- 

Engineering is deduced form the methodical procedure of 

product platform definition. In Figure 5 the basic steps for 

building product releases in terms of Release-Engineering 

are described. 

1. Step: Product functions required by the market and 

related characteristics have to be analysed and clustered 

according to favoured customer segments. By means of 

Release-Engineering the according sales volume for each 

customer segment has to be forecasted considered achieving 

ranking of the most important product features. After 

analysing market demands the customer requirements 

have to be translated to real product features. 

2. Step: Product features are assigned to modules. The 

major challenge is to isolate the most volatile product 

characteristics – so called complexity drivers – on only 

minor modules. Especially volatile product features should 

be composed on easily changeable modules. The achievement 

of the second step is a transparent overview of 

“must-change-in-any-case” modules in the future, to be 

considered in upcoming releases. The focus of the classical 

complexity managements is on determination of common 

and non-common shares. With the focus on Release- 

Engineering, the focus is also on the timewise static and 

dynamic differentiation of product functions. As mentioned 

in the previous section, with the VMEA methodology 

the dynamic evolution of variants - due to emerging 

demands of feature level - over the product lifecycle is the 

most important challenge. Therefore an evaluation matrix 

of the most dynamic features has to be developed. Based 

T1 

T2 

T3 

T4 

T5 

T6 

T7 

T8 

T9 

T10 

T11 

T12 

T13 

T14

on such a dynamic feature, a ranking of release cycle 

predictions and matters has to be made. 

3. Step: Analysis of emerging module interfaces. The main 

challenge is to generate a modular product structure, so 

that the product features influencing complexity most are 

isolated preferably on only a few modules. Main target is 

to create as few as possible module interfaces. 

4. Step: Analysis and optimisation of commonality on part 

level. As mentioned before, high degrees of commonality 

could be achieved, when the product development is based 

on common design rules. The optimisation thereby may 

not only focus on common parts; in order to improve R&D 

effectiveness it is also possible to implicate parametric 

design procedures. Main target is to use the related R&D 

efforts for optimisation of ideal commonality. Likewise 

mentioned in the second step, it will be necessary to comprehend 

and to manage the degree of commonality on part 

level as an active control element for the Release- 

Engineering approach. As seen in Figure 2, a continuous 

adjustment of product variance will be possible at major 

release steps. This major step results in certain decision to 

reduce complexity on part level and to introduce a new 

design layout within the next substantial release. 

5. Step: Analysis and optimisation of part commonality by 

using variant flexible production facilities. Production 

facilities are most suitable when highly variant as well as 

change over flexibility is provided. 

Markets 

Applications 1 

• Basic requirements 

• Specific customer 

and project 

6 

requirements 

Price Quality 

2 

Technical 

Functions 

(concept 

unspecific 

Project unspecific 

Project specific 

Implementation 

3 

Modules 

Modules 

Figure 5: Methodical procedure of release building 

Evaluation of dynamics 

x 

x 

x 

Product- 

4 programme 

Communalities 

Project x 

Project 2 

Project 1 

6. Step: Analysis of ideal clustering of price valuable 

features, especially when those features are highly instable 

over the product lifecycle. Ambition of an ideal clustering 

is to meet customer’s preferences with the offered product 

features and contemporaneously to isolate instable product 

features for minimised internal product complexity. 

The last three steps, the optimisation procedure for achieving 

high degrees of commonality on part and production 

level as well as defining most differential functions in the 

product structure could only be done in an iterative procedure. 

The challenge on the one hand is to define low internal 

product complexity on a static level and on the other 

hand to optimise the commonality degree on product 

feature level over the product programme lifecycle. 

Concluding reflection 

Release-Engineering reduces R&D complexity and increases 

the competitiveness of OEMs and their suppliers 

by adopting this development principle from software 

engineering and introducing it to the field of mechatronics 

Processes 

5 

40 

engineering. R&D efforts can be reduced over the entire 

product programme life cycle. Another effect of Release- 

Engineering is the well-directed control of a product’s 

perceived innovation level in the market place. The process 

of building Release-units needs to be integrated with 

other processes of structuring a product such as modularisation. 

Only if these approaches will intertwine to a large 

extent, cost-effective R&D procedures and a refinancing 

of R&D expenditure is enabled. 

The process of complexity management by managing 

product programme variety on component level has not 

received much research attention yet. In order to fully 

utilise the benefits of Release-Engineering, principles of 

calculation and optimisation of possible product structure 

scenarios have to be aligned with presented programme 

planning by means of Release-Engineering. Along with 

this, the impact of the controlling within the supply chain 

by means of degrees of product maturity for different 

release stages needs to be examined. 

1. References 

1 Schuh, G.: Complexity-Management Study 2004. Laboratory 

for Machine Tools and Production Engineering (WZL) 

and GPS Schuh & Co. GmbH, Aachen, 2004 

2 Clark, K. B.; Fujimoto, T.: Product Development Performance 

– Strategy, Organization and Management in the World 

Auto Industry. Boston, Mass., Harvard Business School 

Press, 1991 

3 Nobeoka, K.; Cusumano, M.: Multi-Project Strategy and 

Market-Share Growth: The Benefits of Rapid Design Transfer 

in New Product Development. MIT Sloan School of 

Management Working Paper, WP #3686-94/BPS, 1994 

4 Womack, J.; Jones, D.; Roos, D.: The Machine that Changed 

the World. Rawson Associates, New York 1990 

5 Cusumano, M. A.: Japan’s Software Factories: A Challenge to U.S. 

Management. Oxford University Press, New York 1991 

6 von Braun, C. F.: The Acceleration Trap in the Real World. In: Sloan 

Management Review, summer 1991, pp.43-52 

7 Crawford, C. M.: The Hidden Coasts of Accelerated Product 

Development. In: Journal of Product Innovation Management, 

9, 1992, pp. 188-199 

8 Schuh, G.; Schwenk, U.: Produktkomplexität managen – Strategien, 

Methoden, Tools. Hanser, Munich 2001 

9 Cusumano, M.; Selby, R.: Microsoft Secrets. The Free Press, a division 

of Simon & Schuster, New York 1995 

10 Freeman, E.: Building Gargantuan Software, American Scientific, 

Winter 1999, pp.28-31 

11 Richter, G.: Schnelle Autos schnell entwickeln. Automobil- 

Entwicklung, 2000, Nr. 11, S. 45-48 

12 Riedel, H.: Mass Customization - Kundenbedürfnisse und 

Skaleneffekte verbinden. Vortrag bei der 3. Aachener Tagung 

Komplexitätsmanagement, Vaals - Niederlande, 2002 

13 Anderson, D. M.: Agile Product Development for Mass 

Customization, McGraw-Hill, New York, 1998 

14 Fine, C.: Clockspeed – Winning Industry Control in the Age 

of Temporary Advantage, Perseus Books, Reading, MA, 

1998 

15 Kipferler, A.; Monti, R.: How Electronics will Revolutionize 

Innovation in Autos, The Boston Consulting Group, Stuttgart, 

Milan, 2000 

16 Eversheim, W.; Schuh, G.: Standard, individualisiert – individuell, 

in: Reinhard, G., Zäh, M. F. (editor), Marktchance 

Individualisierung, Springer, Berlin, pp. 55-63., 2003 

17 www.gps-mbh.de: GPS Schuh Komplexitätsmanagement, 

September 2004

Costing Issues Regarding Product Variant Design 

P. Baguley 1 , D. Schaefer 1 , and Tom Page 2 

1 Design and Manufacturing Research Group, School of Engineering 

University of Durham, 2 Department of Design and Technology, Loughborough University 

Abstract 

Competition for most companies means there is a need to offer their products in a number of different 

variants to their customers. Tools such as computer-aided product configuration systems to automatically 

compose existing components to new variants of a specific product are widely used today. However, in 

contemporary CAE tools there is a lack of intelligent mechanisms to consider such aspects, especially in 

relation to interaction between ECAD and MCAD affects. This paper proposes the use of type-2 fuzzy 

logic to circumvent such problems in capturing the cost of interactions using Integrated Product Teams. 

Keywords: 

Costing, optimisation, type-2 fuzzy logic, automation, product configuration 


ECAD (Electrical Computer Aided Design) is an 

application of design of electrical equipment. The design 

of electrical equipment impacts mechanical design of 

associated components, such as control/switch cabinets, 

etc. The scope of applications of the interaction between 

ECAD and MCAD (Mechanical Computer Aided Design) 

is huge, notably factory design and the design of systems 

that manufacture products being on the large scale side, 

while switch cabinets etc. are on the small scale side. 

An important issue is the integration of ECAD and MCAD 

systems within an overall engineering environment. 

Usually, the electrical and mechanical parts of 

electromechanical components are designed individually 

with separate systems. There is no bidirectional 

modelling incorporating updates on each side. Today, 

existing multi- or interdisciplinary constraints between 

electrical and mechanical design parts cannot be 

modelled properly. Hence, there is no bidirectional 

constraint propagation etc. The clear complexity of this 

interaction impacts the complexity of cost models at all 

stages of the life cycle. 

Product configuration and configuration based modelling 

is a widespread means for the development of product 

variants in MCAD. Recently, this methodology has been 

applied to ECAD as well. In ECAD, configuration objects 

are for instance schematics and diagrams representing 

particular functional units of an electrical installation. 

Plans and diagrams refer to parts that may appear on a 

Bill Of Materials (BOM). Cost knowledge is required for 

entire installations and all the components the installation 

is made of, such as assemblies, sub-assemblies, and so 

on. No previous work has dealt with cost models for the 

ECAD and MCAD interaction but previous cost models 

have appeared separately for ECAD and MCAD ([1]). An 

encompassing cost model including both ECAD and 

MCAD, i.e. an overall systems model, might give 

management more opportunities for optimisation and is a 

growing concern in the design of complex systems. 

41 

Therefore it is important for ECAD to have an indicator 

that communicates the cost of decisions from MCAD and 

vice versa. Because no previous models have been 

made in this way, there is also no previous data 

supporting these types of models. Fuzzy logic is a 

method which can model quantities using no data, but 

experts. Fuzzy logic can be categorised into Type-1 and 

Type-2 fuzzy logic. Type-1 fuzzy logic includes precise 

degrees of membership to a set, whereas type-2 fuzzy 

logic includes degrees of membership to a set that are 

type-1 fuzzy sets themselves. There are very few 

applications of Type-2 fuzzy logic, despite some clear 

advantages in modelling accuracy and robustness found 

by [2]. Type-2 fuzzy logic can be clearly distinguished by 

its efficacy in a team environment. An application of 

ECAD and type-2 fuzzy logic can be integrated 

electronically via spreadsheet or MATLAB files using a 

Product Data Management (PDM) system with the 

Computer Aided Design system. 

2 EXAMPLE PROBLEM 

A company wants to develop an elevator that can carry 

10 persons. The elevator has electrical parts as well as 

mechanical parts and there exist a lot of constraints 

between them. However, division A of the company 

develops the mechanical parts with an MCAD system 

and division B develops the electrical ones with an ECAD 

system. There is no integration of both systems. 

Therefore, the corresponding engineers talk to each 

other in a conventional sense and then continue working 

with their CAD systems. There is no interdisciplinary 

concurrent engineering at all, neither in respect to the 

general engineering nor to costing aspects relevant to 

management, controlling and other decisions of the 

company. 

Some months later, the same company is asked to 

develop two variants of their elevators, one for carrying a 

maximum of 5 persons, one for a maximum of 25 

persons. The problem is to develop the two variants 

based on existing drawings, plans, etc., and to calculate 

costs.

3 COST MODELLING TYPES 

There are a few main types of cost models that are 

associated with particular stages of the product 

development cycle. The 2 main examples of this are: 

• Grass roots cost models, and 

• Parametric cost models ([3]). 

Grass roots cost models break down a product into a 

very detailed Bill Of Materials (BOM) and associated 

detailed activities. Grass roots cost modelling suffers 

from resource intensive data collection that can only 

occur late in the design stage when the detail required is 

available. In addition, if the detail is available, the large 

amount of summations involved of the small details can 

become a significant error in cost. Parametric cost 

models are those most significant to this research. 

Parametric cost models relate product specifications to 

cost. These product specifications can be geometric or 

functional or other, and typically occur early in the design, 

for example concept or aggregate (as defined by [4]). 

Parametric cost models produce relationships between 

the individual costs of previous variants for which cost 

estimates exist, and their specifications, identified as 

“cost drivers”. Such relationships are termed Cost 

Estimating Relationships (CERs) and are typically 

produced using regression analysis. 

4 DATA SOURCES 

A key part of cost modelling is in identifying and 

collecting relevant data from data sources, the part that is 

the key constraint in subsequent cost modelling tasks. 

The lack of data informing designers of the bidirectional 

impact of MCAD with ECAD means cost modelling 

methods that can operate with no data are essential. 

Fuzzy logic is one such method that can use expert 

opinion and intuition to form models. Fuzzy logic that can 

be used by an Integrated Product Team (IPT), i.e one 

that can lever expertise in both ECAD and MCAD, is 

Type-2 fuzzy logic. 

5 SOLUTION TO THE PROBLEM 

The proposed solution to the problem is to capture the 

interaction between ECAD and MCAD using expert rules. 

Such expert rules can be based on the dependencies 

between design entities, a requirement of parametric 

modelling in electrical design identified by [5]. With the 

growing standardisation of parametric parts in both 

ECAD and MCAD (known as the “box of bricks” in 

MCAD) a more general set of expert rules can be 

foreseen applied to the general problem of the interaction 

between standard parts. For our elevator problem a 

sample of the set of rules could be: 

If “Number of People” is “Large” then “Motor” is “Medium” 

If “Motor” is “Medium” then “Motor Cost” is “Small” 

If “Motor” is “Medium” then “Cable Strength” is “High” 

If “Cable Strength” is “High” then “Cable Cost” is “Very 

High” 

The interactions are captured as expert knowledge in a 

series of structured look up tables. It is envisaged in the 

future, when the method will be a mature one, that such 

tables can be supplied by the manufacturers of electrical 

equipment. Hence the knowledge becomes a marketable 

product. This ultimate knowledge capture mechanism 

within fuzzy logic can be termed that of a grammar ([6]). 

6 FUZZY SETS 

A central issue surrounding fuzzy logic is that of fuzzy 

sets. In conventional Boolean thinking elements can 

belong to a set (represented by the value 1), or not 

belong to the set (represented by 0). The boundary is 

very well defined. Fuzzy sets are different in that 

elements can partially belong to a set to a degree 

between 0 and 1. Hence a person can belong to the set 

of tall people (to a degree 0.3) and also to the set of short 

people (to a degree 0.3). It is noticed that the degrees of 

membership to the sets of tall people and short people do 

not necessarily add up to 1. The concepts of tall and 

short are fuzzy. Fuzzy sets were developed by Zadeh 

([7]) because humans do not think in a Boolean sense, 

so that there is not a well defined barrier between being 

short and being tall, and in fact any calculations based on 

a well defined barrier should come under question. Two 

fuzzy sets are shown in Figure 1, where it is noticed that 

a person is only thought of as certainly tall (degree of 

membership 1 to the fuzzy set, tall) if they are 6 foot and 

over. Of course this is subjective and might be moved 

higher or lower, or other. 

Figure 1: Type-1 fuzzy sets 

7 TYPE-2 FUZZY LOGIC 

Type-2 fuzzy logic takes the concept of fuzzy sets further 

by recognising that degrees of membership to a fuzzy set 

are also precisely defined. Mendel ([2]) and John ([8]) 

describe how uncertainty is not fully captured by precise 

degrees of membership since precision implies a form of 

certainty. Hence type-2 fuzzy logic should be considered 

as a system when the degrees of membership to a set 

are fuzzy sets themselves. In addition, when utilising 

type-2 fuzzy logic, there is the process of type reduction 

to reduce type-2 fuzzy sets to type-1 fuzzy sets after 

performing inferencing. (This is an algorithmic concern 

but also has connotations for the interpretation of 

uncertainty). Subsequently it is possible to defuzzify the 

type-1 (reduced type-2) fuzzy set. John ([8]), provides a 

formal definition from Karnik and Mendel ([9]), "a type-2 

fuzzy set is characterised by a fuzzy membership 

function, i.e. the membership value (or membership 

grade) for each element of this set is a fuzzy set in [0,1], 

unlike a type-1 fuzzy set where the membership grade is 

a crisp number in [0,1]". 

8 SEQUENCE OF STEPS IN PRODUCING TYPE-2 

FUZZY LOGIC MODELS 

The sequence of steps in producing type-2 fuzzy logic 

models can be summarised (as shown in Figure 2): 

1. identify the “concepts” to be used in the model 

2. Identify expert rules 

3. Produce type-2 fuzzy sets for the concepts in the 

expert rules using knowledge acquisition methods, 

or newly developed automatic methods ([10]) 

42

4. Identify fuzzy logic operators by consensus from the 

existing literature, or by using a decision making 

methodology. A decision making methodology was 

developed by Baguley and Stockton (2002) within 

the EPSRC Grant Reference: GR/M58818/01, "IMI: 

IMPROVING THE COST MODEL DEVELOPMENT 

PROCESS (COSTMOD)", for the choice of type-1 

fuzzy logic structural elements. 

5. Refine choices in (1), (2), and (3) through a learning 

mechanism. 

Figure 2: Sequence of steps in using fuzzy logic 

9 IDENTIFY CONCEPTS IN THE MODEL 

Concepts in the model include the nouns and adjectives 

of the problem. For example the nouns, “motor”, “lift”, 

“time”, “labour”, “cost”; and the adjectives “low”, 

“medium”, “high”, and “very”. Brain storming methods are 

typical for identification, for example the use of 

relationship diagrams, as shown in Figure 3, allows the 

picturing of concepts and the relationships between them 

using arrows. The formation of rules is made easier using 

such diagrams, i.e. the identification of relationships 

using the arrows. 

Number of People 

Lift Weight 

Depends on 

Lift Size 

Motor Power 

Concept 

Relationship 

Figure 3: Relationship diagram to identify concepts for a 

fuzzy logic model 

10 IDENTIFY EXPERT RULES 

The format of expert rules (if then rules) is familiar to 

many. For example if “size” of “lift” is “medium” then 

“motor” “power” is “low”. Of importance is whether the 

rule base is to be a “complete” rule base or an informal 

and ad-hoc collection of rules the expert uses whilst 

informally performing trade-offs between cost and 

changes in MCAD and ECAD designs. A complete rule 

base involves every single combination of variables. 

Complete rule bases consume more resources to build, 

and when there are lots of rules, are difficult to form 

effectively. 

43 

11 PRODUCE TYPE-2 FUZZY SETS 

Firstly the range of the variables (also known as the 

Universe of Discourse) with which to cover with fuzzy 

sets, must be identified. Subsequently the following 

decisions are required: 

1. shape of the fuzzy sets (do not all have to be the 

same, for example a mixture of triangular and 

trapezoidal shapes) 

2. parameters of the shape of the fuzzy sets (for 

example the width and centre point of a triangular 

fuzzy set) 

3. overlap between fuzzy sets (for example how much 

fuzzy set widths coincide on the range of the 

individual variables) 

4. “blurring” of the fuzzy sets, i.e. making the type-1 

fuzzy sets into type-2 fuzzy sets (as shown in Figure 

4), (for example each precise degrees of 

membership to a fuzzy set are made into a type-1 

fuzzy set) 

This sequence of tasks, here, is not a unique one. The 

decision can be a difficult one, but is aided by the robust 

properties of fuzzy logic, i.e. errors in detail of the shapes 

of the fuzzy sets can be accommodated by the model in 

that accuracy is not greatly reduced. In addition, there is 

the potential extra robust properties of type-2 fuzzy sets. 

Degrees of membership to a fuzzy set and the ideas of 

set theory are essential, but work has been done in which 

such knowledge is shielded from experts by indirect 

questions provided in a written questionnaire ([11]). 

Therefore only their domain knowledge is required. 

Figure 4: Type-2 fuzzy set 

12 FUZZY LOGIC OPERATORS 

The fuzzy logic operators are mathematical operators or 

rules for combining fuzzy sets for inferencing and 

particularly involve: 

(1) connectives in rules, e.g. the “AND” operator 

(2) implication operator for using rules 

(3) aggregation operator for combining the results of 

rules 

13 TYPE-REDUCTION 

Type reduction involves the conversion of the type-2 

fuzzy set produced after inferencing into a type-1 fuzzy 

set. 

14 DEFUZZIFICATION 

Defuzzification converts a fuzzy set into a single value. 

This step is most appropriate for control system 

applications.

15 REQUIREMENTS FOR A KNOWLEDGE 

ACQUISITION METHOD FOR TYPE-2 FUZZY 

LOGIC AND ECAD 

An essential step in using fuzzy logic as proposed in this 

research is the knowledge acquisition phase. Experts' 

domain knowledge is mapped into fuzzy logic structural 

elements. Some of the requirements of a knowledge 

acquisition method are shown below. 

1. To capture the terminology of ECAD and MCAD 

engineers 

2. To consistently capture the parameters of type-2 

fuzzy sets using (1) 

3. To hide fuzzy logic knowledge from ECAD and 

MCAD experts 

16 DIGITAL ENTERPRISE TECHNOLOGY (DET) 

ARCHITECTURE FOR THE USE OF TYPE-2 FUZZY 

LOGIC AND ECAD. 

Due to the lack of research and practice into type-2 fuzzy 

sets, there are no commercial software packages 

available to perform type-2 operations. Mendel ([2]) at the 

University of California has produced “m-files” that can 

perform some type-2 operations, for the MATLAB fuzzy 

logic toolbox. The MATLAB fuzzy logic toolbox is a well 

known software package for performing type-1 fuzzy logic 

operations. The essential aspect of MATLAB is its 

intuitive Graphical User Interface. The GUI promotes the 

modelling of expert intuition. Digital Enterprise 

Technology (DET) is defined as: "the collection of 

systems and methods for the digital modelling of the 

global product development and realisation process in 

the context of life cycle management". A fuzzy logic 

model can depart from an integrated systems 

architecture, by standing outside of any collection of 

software components, and to be used by an expert to 

express his opinion (Figure 5). The expert may use 

existing DET components within a DET architecture to 

formulate this opinion, but need not connect MATLAB to 

any data sources, as would be the case in processing 

numerical data. 

Distributed and Collaborative 

Product Design 

Distributed and Collaborative 

Process Design & Planning 

Equipment and Plant Layout 

Design & Modelling 

Physical-to-Digital Environment 

Integrators 

Technologies for Enterprise 

Integration & Logistics 

Figure 5: Observer using a DET architecture to formulate 

opinion 

17 HOW TO USE THE TYPE-2 FUZZY LOGIC MODEL 

1. Crisp values (non-fuzzy, i.e. ordinary numbers) for 

the input values are put into the model, 

2. Because each value corresponds to a degree of 

membership to a type-2 fuzzy set, which is a type-1 

fuzzy set, the input value corresponds to this type-1 

fuzzy set, 

3. This type-1 fuzzy set is combined with similar type-1 

fuzzy sets produced for the other input variables, 

P 

D 

M 

from their respective input values, as indicated by 

each rule in turn, 

4. The resulting type-1 fuzzy set from the several 

combinations of input values, is combined with the 

type-2 fuzzy set corresponding to the output variable 

for each rule in turn, 

5. All these resulting type-2 fuzzy sets from each rule 

are aggregated together to form one overall type-2 

fuzzy set, 

6. this type-2 fuzzy set undergoes a type-reduction 

operation to form a type-1 fuzzy set, 

7. this final type-1 fuzzy set is defuzzified. 

The structure of a type-2 Fuzzy Logic System is shown in 

Figure 6, It is clearly seen that there is a separate typereduction 

stage, this is needed to defuzzify type-2 fuzzy 

sets. 

Figure 6: Type-2 fuzzy inference system 

18 CONCLUSION 

There is a gap in provision for modelling the design of 

electrical and mechanical components when they affect 

each other. The interaction of electrical and mechanical 

components, for example when predicting costs, is 

captured using expert rules and quantified using type-2 

fuzzy logic. Type-2 fuzzy logic is used because of the 

exceptional level of uncertainty within an Integrated 

Product Team. This uncertainty is especially prevalent 

because of the diversity of the 2 interacting areas of 

ECAD and MCAD. 

19 REFERENCES 

[1] Aguirre, E., and Raucent, B. (1994), Economic 

Comparison of Wire Harness Assembly Systems, 

Journal of Manufacturing Systems, Vol 13 No. 4., 

pp. 276-288. 

[2] Mendel, J. (2001), Uncertain Rule-based Fuzzy 

Logic Systems, Prentice Hall, ISBN 0-13-040969-3. 

[3] Rush, C. and Roy, R. (2001), Expert Judgement in 

Cost Estimating: Modelling the Reasoning Process, 

Concurrent Engineering: Research and 

Applications, Vol. 9, No. 4., pp. 271-284. 

[4] Maropoulos et al (1998), CAPABLE: an Aggregate 

Process Planning System for Integrated Product 

Development, Journal of Materials Processing 

Technology 76, pp.16-22. 

[5] Schaefer, D., Eck, O., and Roller, D., A Shared 

Knowledge Base for Interdisciplinary Parametric 

Product Data Models in CAD, In: Lindemann, U., 

Birkhofer, H., Meerkamm, H., Vajna, S. (Eds.): 

Proceedings of the 12th International Conference 

on Engineering Design: ICED ’99, Volume 3. 

Garching: Technische Universität Muenchen, 1999. 

- Munich, Germany, August 24-26, 1999. - ISBN 3- 

922979-53-X, pp. 1593-1598. 

[6] Hsiao, S., and Chen, C. (1997), A Semantic and 

Shape Grammar Based Approach for Product 

Design, Design Studies 18, pp. 275-296. 

[7] Zadeh, L.A. (1965), Fuzzy Sets, Information and 

Control, 8, pp. 338-353. 

44

[8] John, R.I. (2002), Embedded Interval Valued Fuzzy 

Sets, Proceedings of the Fuzz-IEEE 2002, pp. 

1316-1321. 

[9] Karnik, N.N., and Mendel, J. (1998), An Introduction 

to Type-2 Fuzzy Logic Systems, Technical Report, 

University of Southern California, 1998. 

[10] John, R.I., and Czarnicki, C. (1999), An Adaptive 

Type-2 Fuzzy System for Learning Linguistic 

Membership Grades, IEEE International Fuzzy 

Systems Conference Proceedings, August 22-25, 

1999, Seoul, Korea. 

[11] Arnold, F., Moody, N., Reiter, W.F., Maunder, C., 

Rogers, B., Baguley, P., Chapman, P., Lomas, C., 

Zhang, D., and Maropoulos, P. (2004), An Extended 

Virtual Enterprise SMARTEAM Engineering Project, 

2 nd International Seminar on Digital Enterprise 

Technology, Seattle, Washington, USA, September 

13-15 th 2004. 

45

A New Method for Variant Design Technology in ECAD 

D. Schaefer 

School of Engineering, University of Durham, UK 

Abstract 

One of the most important approaches to enable cost and time reduction with respect to Computer-Aided 

Design for electrical/electromechanical engineering (ECAD/ECAE) is to develop, generate, and handle design 

variants in an efficient manner. The objective of this paper is to present a generic variant design technology 

approach that has a high potential for the efficient reusability of existing projects. More precisely, the paper 

will present the procedure to allow the new methodology in variant design technology to be implemented 

within an arbitrary ECAD environment. The method presented automatically generates a complete technical 

documentation of an electrical installation on the basis of a placed order specification. This involves three 

major steps. Firstly, a product variant of an installation is configured on the basis of existing standardised 

modules. Secondly, based upon the corresponding configuration file, a set of commands describing the 

generation of a typical ECAD project containing the configured modules is automatically compiled. This is a 

key novelty, as all commands are expressed in a non-system specific meta-language, which can then be 

automatically translated into a macro programming language of a specific ECAD system and stored as a file. 

Thirdly, the specific ECAD system can import and process the file to create a practical ECAD project of 

realistic complexity. 

Keywords: 

Variant Design, ECAD/ECAE, Process Automation, Product Configuration 


In order to remain competitive in the market, most 

companies offer their products in a range of different 

variants. However, new variants of existing products are 

more and more often composed of existing basic 

components rather than newly designed. A precondition to 

allow for this is to have modular product structures. An 

industrial branch where this is common practice is plant 

engineering and construction. The majority of companies 

developing electrical/electromechanical installations tend 

to reuse their existing designs, plans and project 

documentations to develop additional customer specific 

variants. 

Unfortunately, this process is predominantly performed 

manually, which is far from being efficient and effective. 

Due to the permanently increasing pressure of competition 

it is vital to develop automatisms that facilitate the 

generation of product variants. 

The basic idea behind the variant design approach 

presented in this paper is simply to automate the workflow 

process of creating ECAD variants as it is manually 

performed by most companies today. The approach is 

closely related to industrial best practice and derived from 

day-to-day operations. 

1.1 The basic concept as applied in practice 

Many companies approach their potential customers by 

technical field sales and distribution staff. These sales 

persons usually have selling catalogues on their disposal, 

which allow customers to assemble bespoke product 

variants based on basic modules or components that can 

be combined. Hereby, the number of components that can 

46 

be chosen from (at this stage of the sales process) tends 

to be relatively small. 

Once a rough pre-configuration is finished, design 

engineers and other office employees get together to 

determine a resultant technical fine configuration. This fine 

configuration comprises all parts and components 

necessary to make up the product variant desired and – 

depending on the nature of the modular product structure 

associated – may consist of thousands of items. This fine 

configuration also includes technical knowledge in terms of 

rules and constraints pinpointing under what 

circumstances what components may or may not be 

combined with others. 

In order to identify specific components or items, designers 

usually refer to part numbers, drawing numbers, BOM 

entries, or similar terms. Specifically in the area of 

electrical/electromechanical engineering most of these 

configuration items are schematic diagrams, terminal plans 

or a variety of other documents required to express 

functionality and assembly of an installation. 

The next step in creating the configured variant requires a 

design engineer to start an ECAD system, generate a new 

project, copy the plans configured into the project and 

specify individual customer and order details. 

Subsequently, alterations necessary to the specific project 

have to be accomplished and the process of generating an 

updated complete electrical documentation according to 

the alterations or amendments made has to be initiated. 

In order to make the development of product variants more 

effective, ECAD system vendors aim towards an automatic 

computer-aided support of the workflow process outlined 

above.

2 VARIATIONAL DESIGN TECHNOLOGY APPROACH 

2.1 Functional Principle 

The basic functional principle of the variant design 

approach presented in this paper is now described. It is 

based on aspects from knowledge-based product 

configuration, the programming of variants, as well as 

parametric modelling and process automation. The 

fundamental idea behind the approach is to automatically 

generate an entire technical documentation of an 

electrical/electromechanical installation on the basis of a 

placed order specification. The overall process to achieve 

this involves five steps as shown in figure 1. 

Identification of standardised components of 

an electrical installation to represent a 

modular structured product variant 

Composition of identidied components in 

terms of data structure expressions 

Automatic compilation of commands in non 

system specific meta-language describing the 

generation of a typical ECAD project containing 

the previously configured modules 

Automatic translation of meta-language 

commands into a macro programming 

language of a specific ECAD system 

Import and processing of generated file to 

create practical ECAD project 

Figure 1: Functional principle of 

ECAD variant design technology approach. 

Firstly, based upon a modular product structure a variant of 

an installation is configured by composing standardised 

modules of existing (previously developed) components. 

Secondly, all components identified to compose a specific 

variant configuration are stored in a data file. This includes 

settings for parameter values that may finally define some 

characteristics of the components. This configuration file 

corresponds to a bespoke data structure newly developed 

and appropriate to describe ECAD product variants in 

general. 

Thirdly, based upon the configuration file created, a set of 

commands describing the generation of a typical ECAD 

project containing all the components configured together 

is automatically compiled. This is a key novelty, as all 

1. 

2. 

3. 

4. 

5. 

commands to generate such a project are expressed in a 

non system specific meta-language. 

Fourthly, the variant project description expressed in non 

system specific meta-language commands is automatically 

transformed into a batch file of commands of a macro 

programming language of associated to a particular ECAD 

system. 

Fifthly, the specific ECAD system imports and processes 

this batch file to create an actual ECAD project. This 

project then may be handled or dealt with in any way the 

ECAD system allows. 

2.2 Technology Approach 

A technical approach to realise the functional principle is 

now given. An overview is illustrated in figure 2. Since step 

one of the functional principle is obvious, step two is the 

point to start with going into further detail. 

As already mentioned in paragraph 2.1, a data structure 

specifically tailored for representing configuration based 

variant projects is required. This data structure primarily 

has to cover components and documents typically used to 

make up an entire project documentation of an electrical 

installation. Furthermore, the data structure it has to span 

semantic information on relations and constraints existing 

between components or documents. In order to be able to 

store, import, process and automatically evaluate variant 

projects based on a semantics-based data structure in 

subsequent processes, a standardised data format for 

representing the data structure has to be chosen. In the 

approach described in this paper the variant project data 

structure has been expressed in terms of a document type 

definition (DTD) associated to the language XML 

(extensible mark-up language). Simply speaking, XML is a 

mark-up language for describing hierarchically composed 

objects that strictly distinguishes between structure, 

content and layout of the objects data. The specific 

purpose of a DTD is to define the structure of XML files. 

Based upon these technical arrangements, a variant 

project expressed in terms of XML can be validated 

against a DTD to check whether or not its structure 

complies with the requirements defined in the DTD. So far, 

configuration based variant projects (components, 

documents, relations, constraints, etc.) whose original 

underlying data structure has been transferred into a XML 

pendant can be stored as XML data files to be further 

processed later on. 

Realising step three of the functional principle involves 

formulating commands expressing the generation of a 

configuration based ECAD project in a non system specific 

form. Hence, a suitable meta-language has been 

developed and applied. This meta-language covers a 

variety of system commands similar to those being typical 

for contemporary ECAD system’s macro languages. 

Following the functional principle of paragraph 2.1, a 

variant project stored as XML file now can be transferred 

into a new data format describing the generation of an 

ECAD project containing the configured components. 

Technically speaking, this means to enhance the original 

XML data structure of a configuration based variant project 

in such a way that it allows to model and express 

commands in the meta-variant-construction language ‘vcl’. 

In order to allow for further standardised data processing 

operations to be performed on vcl files, the data structure 

of the meta-language vcl has to be transferred into an 

equivalent XML DTD. Henceforth, this meta-language 

expressed in terms of an XML DTD will be denoted as 

variant construction language in XML, in short ‘vclX’. To 

sum-up, the components of a configured ECAD variant 

project stored as XML file have been picked-up and 

transferred into another, more sophisticated and powerful 

47

New Method for ECAD Variant Design Technology 

ECAD Variant Project 

data structure 

of 

configuration based 

variant project 

transfer given data 

structure into XML-DTD 

to allow standardised 

data exchange 

store variant 

project as 

XML-file 

Transformation of data 

structure into vclX 


enhanced data structure 

of variant project 

containing commands for 

ECAD project generation 

in meta-language vcl 

transfer vcl into XML- 

DTD (vclX) to allow 

standardised data 

exchange 

store variant project 

containing project 

generation commands 

as XML file 

Transformations are described in terms of XSL files 

Automated transformations are realised by 

processing given XSL files using XSLT 

XML data structure containing non system specific 

commands to describe the generation of an ECAD project 

made up of the components configured. 

The above transformation process is carried out 

automatically using ‘XSL’ (Extensible Style sheet 

Language) and a software tool called ‘XSLT’ (XSL 

Transformation). XSL is a language specifically developed 

to facilitate transformation purposes and allows defining 

rules describing the transformation from one XML structure 

into another data structure. The transformation rules 

necessary to perform the desired transformation are stored 

within a specific XSL data file. The actual data 

transformation then is carried out by XSLT which requires 

three things to work: (1) a source XML file to be 

transferred, (2) the source file’s corresponding DTD and 

(3) a XSL file containing all the rules describing the 

mapping onto the target XML data structure. 

The procedure described above analogously recurs for 

step 4 of the technology approach. However, this time an 

XSL file describing the transformation from the non-system 

specific vclX command list structure into another structure 

encompassing commands of a specific ECAD system’s 

macro language is required. The result of this final 

transformation is a batch file to be imported and processed 

by the specific ECAD system chosen. In other words, a 

real ECAD project in a native data format has been 

created. 

2.3 Module Conception 

In this paragraph a module conception for implementing 

the approach discussed above is presented. It is tailored to 

support a workflow closely related to that described in 

Transformation of vclX into 

ECAD macro language 


ECAD data structure of a 

variant project containing 

commands for ECAD 

project generation in 



as data exchange 

format of an ECADsystem 

Store variant project in ECAD 

macro language or.native data 

format of ECAD system 

Figure 2: Approach to an automatic generation of ECAD project 

48 

paragraph 1.1. For an overall illustration of the module 

conception see figure 3. 

The variant module basically comprises of two submodules, 

notably the ‘configuration module’ and the 

‘coupling module’. It is developed as a self-contained unit 

that may be coupled to one or more ECAD systems rather 

than directly implemented within a specific system. 

The purpose of the configuration module is to perform the 

tasks carried out by design engineers and office 

employees as described in paragraph 1.1. This means to 

either create new configurations or adjust existing ones 

based upon a modular product structure using existing 

basic parts or components. Hereby, the various plans of an 

electrical documentation (e.g. schematics, terminal plans, 

part lists, etc.) are drawn on as configuration objects. 

Prior to being able to work with the configuration module all 

the components (projects, sub-projects, etc.) existing on 

ECAD site and meant to be available for variant projects 

have to be imported to the configuration module’s 

database. 

Using the features of a comfortable graphical user 

interface (GUI), the imported ECAD components can be 

used to combine new variant projects, to adjust previously 

created variants and to store further basic components into 

the database. The design knowledge regarding rules 

describing possible combinations and configurations has to 

be brought into the database as well. The system kernel of 

the configuration module therefore has to incorporate an 

intelligent mechanism to maintain, check and control the 

compliance of constraints modelled.

ECAD Variant Module 

Configurator 

XML data 

model 

Kernel 

manual import of projects, components and other data 

Coupling module 

Kernel 

database database 

Figure 3: System architecture of the variant module 

Due to the complexity of knowledge based configuration 

systems the development of proprietary knowledge based 

configuration tools is not recommended. There are many 

commercial solutions for almost any configuration tasks 

available on the market. 

The purpose of the coupling module is to automate those 

steps of the workflow process that today are usually 

Non 

system 

specific 

language 

Figure 4: Example of a simple schematic automatically generated 

with the variant module 

Interface sets 

Macro 

language 1 

Macro 

language 2 

Macro 

language 3 

Macro 

language n 

performed manually by an engineer after the components 

for a variant configuration have been determined. 

A first task for the coupling module is to import a variant 

configuration from the configuration module. Subsequently, 

it has to create a batch file of non-system specific 

commands describing the relevant steps to open a new 

ECAD project and to include the chosen components. After 

that, the coupling module has to transfer 

this non-specific commands into a batch 

file for a specific ECAD system which can 

import and process the final batch file 

then. 

3 CONCLUSION 

The approach presented in this paper has 

been realised as a software prototype. Its 

general applicability has been proven and 

several possibilities to amend, extend, or 

standardize the suggested technological 

approach are currently being investigated. 

Commercially enhanced, a variant module 

like the one proposed in this paper would 

bear a high potential in respect to cost and 

time reduction. 

4 REFERENCES 

[1] Schaefer, D., 2003, Variantentechnologie 

unter besonderer Berücksichtigung 

von Elektrotechnik CAD, 

Shaker-Verlag, Aachen, ISBN 3- 

8322-1901-3, 292 pages 

49

PART 3 

MICRO ELECTRO-MECHANICAL 

SYSTEMS (MEMS) 

50

Use of CoventorWare in the Design of Innovative IC Probes Based on 

Microsystems Technology 

M.D. Cooke and D. Wood 

Microsystems Technology Group, School of Engineering 

University of Durham, Durham, England DH1 3LE 

Abstract 

This paper presents a detailed examination of the use of the CoventorWare Tm software in the design of 

novel IC probes. Analysis is made of the factors which can influence the device reliability and 

performance. Discussion is made of the usefulness of the analysis in comparison to actual experimental 

data. We also demonstrate the usefulness of our devices and determine how future designs may be 

further optimised using the software as a design aid. 

Keywords: 

Integrated Circuit Probe Card, microsystems, CoventorWare Tm 


A standard probe card to test integrated circuits (ICs) is 

a piece of precision mechanical engineering. It consists 

of a large number of individually assembled probes, 

usually arranged either in a circular or a rectangular 

perimeter, around a space where the chip will sit. The 

probes are individually addressable electrically, and 

have all to be assembled and set at the same correct 

height on the card in order to function. The complete 

card takes a long time to assemble, and is large and 

expensive. The cost of a probe card is in the range £500 

- £5000, or even higher, depending on complexity. No 

significant breakthrough has been made in probe card 

design for many decades. 

Previous microsystems work on IC probes has been 

limited mainly to silicon micromachining or simple metal 

designs [1-3]. These tend to have issues with contact 

resistance or durability. There are also associated issues 

in producing enough stress to make the metal 

cantilevers all deflect evenly. Recently more complex 

structures have been developed. The design by Kataoka 

et al. was a complex electroplated spring structure 

comprising multiple fabrication stages which could be 

made straight onto a circuit board [4]. Krüger et al. 

demonstrated a beam design optimized on multiple 

anchor points to minimize stress with a self alignment 

feature for contact bumps [5]. 

Membrane 

Substrate 

IC lowered 

Probe Bump 

Self-Limiting 

Sensor 

Support Columns 

Figure 1: Schematic diagram of the IC probe 

We have developed a new approach to fabricating IC 

probe cards based on Microsystems Technology. It 

involves the fabrication of electroplated nickel or gold 

structures that are both flexible and electrically 

51 

addressable. The design has unique features to allow it 

to be highly durable for application in the workplace. 

Our design can be seen in figure 1. The electroplated IC 

probes are designed as a membrane structure with a 

probe bump in the centre. This membrane, which in 

principle can be any geometry, is supported by a number 

of columns with a self-limiting sensor located beneath 

the structure to allow control of the applied force. These 

structures can be mass manufactured in large arrays 

suitable for use as a probe card, and an additional 

benefit is that there is only a minor design change to 

make the probes suitable for either contact pad or solder 

bump designed circuits. Our typical structure is a beam 

design, with a membrane dimension of 1000µm x 100µm 

x 5µm supported on 2µm high columns. Deflection is 

less than 3µm under an applied force of 2mN giving a 

contact resistance below 1?. We have the very real 

promise of cheap flexible probe designs which can be 

produced quickly. In order to optimise our current probe 

devices, the CoventorWare package has been used to 

model their performance under loading as seen when an 

IC pad or bump would make contact. This allows the 

design to be optimised before any fabrication work is 

undertaken. This means that the time in producing the 

samples is significantly reduced, and the work required 

to determine the optimum dimensions, materials etc is 

much less. Also it allows the design solution for a 

particular problem to be produced rapidly. We 

particularly examine the effect of membrane geometry 

and dimensions on the deflection under load. 

2 COVENTORWARE SOFTWARE 

2.1 Overview of Software 

CoventorWare is an integrated microsystems analysis 

environment programmed to determine an effective 

design before experimental work is undertaken [6]. It has 

a number of individual modules which the user can 

specifically purchase. 

The Designer module allows the development of a three 

dimensional model of a complex device. Integrated into 

this is the process emulator which emulates the 

fabrication steps for Microsystems devices. It uses 

standard deposition processes and etching with control 

of the bulk and thin-film geometries.

The layout editor then allows mask design as used in the 

process flow. Integration of a wide range of materials 

allows a more accurate evaluation of true device 

behaviour. This information is then all combined to give 

a 3D generated model, which allows meshing of 

individual components for further analysis. 

The Analyser module is a range of tools specifically 

designed to fully determine how a device will interact 

under specified conditions. It is split into several sub 

packages, for example, MemElectro examines 

electrostatic properties, MemMech steady-state thermomechanical 

properties and CoSolve-EM analyzes 

coupled electromechanics with hysteresis. Further 

packages allow analysis of damping, piezoelectric, 

thermal effects and electrical characterisation. More 

detailed modelling allows the introduction of the package 

and external environment to the system. These results 

can then be viewed using TechPlot, which allows both 

3D and graphical information to be extracted from the 

modelling. 

2.2 Probe Design Specific Information 

To apply our device analysis to the software system, it 

was essential to set up the parameters of our design, in 

the process designer. Initially, we characterised the 

individual device layer thickness and determined the 

fabrication route using the process editor. The defined 

substrate was silicon , with a 1000Å thermal oxide 

layer on the surface to limit any electrical conductivity 

and isolate it from the device. The column, membrane 

and contact regions could then all be specified for their 

individual thickness, with a sacrificial defined layer used 

to fully simulate the device design process. In Table 1 

we show a typical process flow as used for our designs 

(the parameters are varied as the device optimisation is 

examined). 

Table 1: Process flow parameters as used in the 

examination of the IC probe design in the Process 

Designer element of CoventorWare 

The information from the process design stage was then 

integrated into the design layout editor. Using the simple 

graphical interface, it was possible to apply the typical 

design of our structures as a two dimensional image 

viewed from the top side. A typical design in 2D can be 

seen in Fig 2. 

By linking each structure to the information stored in the 

process designer, it was possible to build a three 

dimensional model of the device. 

52 

Figure 2: Typical beam probe design as fabricated using 

the design layout editor 

The next stage in the device development was to use the 

Preprocessor to view and provide the initial modelling of 

the device. The initial design showed all the 

components. For our modelling, it was important to 

define two specific regions, the first being the substrate, 

the second being the device structure. All components 

were defined as solid conductors then added for mesh 

modelling. Before modelling was undertaken, it was 

important to define the key faces which had to be 

referenced for interaction during the modelling. These 

included the anchor point, membrane bottom, substrate 

surface and top of the contact. 

As shown in Fig. 3 for our typical device, region 1 (the 

substrate) was modelled using a surface mesh with 

individual mesh size of 20 microns. The components in 

region 2 (the device) were modelled using an extruded 

brick mesh using a pave algorithm, with element size of 

5 microns. The pre-processor was then used to generate 

the mesh. 

Figure 3: Three Dimensional model of the probe 

structure with appropriate meshing on each individual 

component. Blue corresponds to region 1 (the 

substrate). Green (the support columns), red (the 

membrane beam) and orange (the probe bump) make 

up the device (region 2). 

The final stage in producing the model for the device 

was to use the analyser package. The analysis used for 

measuring the probe deflection was the mechanical 

settings. Surface boundary conditions were applied, and 

the anchor points were fixed. At the top of the probe 

bump, increasing pressure in the z-plane was applied, 

ranging from 0 to 10 Pa. This parametric range was then 

applied to the membrane to give the variation of 

deflection as a function of applied pressure. The typical 

expected deflection as a function of pressure can be

seen in Fig 4, while we show the profile of the beam 

during deflection in Fig 5. As expected the maximum 

deflection is in the centre where the force is applied to 

the probe pump. However the deflection is not linear with 

applied force. 

Deflection (μm) 

3.0 

2.5 

2.0 

1.5 

1.0 

0.5 

0.0 

0 2 4 6 8 10 

Applied Pressure (Pa) 

Figure 4: Model of the deflection of a 1000 x 100 micron 

beam of thickness 5 microns made from nickel on 4 

micron high support columns. 

Figure 5: Modelled deflection profile of the IC probe card 

device at maximum deflection 

It was then possible to examine the data to show a 

number of key parameter variations as a function of the 

applied pressure. For instance the stress in the sample 

may be modelled. This can provide an insight into the 

regions where the device would fail during repeated 

cyclic loading. Also it is essential in the implementation 

of the self limiting sensor as this can then be employed 

to determine how the optimum deflection can be 

achieved. 

Fig 6 shows the deflected cantilever and highlights the 

regions of stress. As expected the main stress regions 

were those found near the column support regions giving 

rise to high stress which could result in the damage of 

the membrane surface. Secondary high regions of stress 

were found at the position where the sample is in contact 

with the underlying surface. 

53 

Figure 6: Modelled stress of the membrane under full 

deflection 

2.3 Specific Analysis for Probe Optimisation 

Using CoventorWare, it was possible to undertake more 

specific analysis of the beam design for use in the probe 

card. We performed the analysis of the change in beam 

deflection for a fixed applied pressure ( 0.25 Pa). In Fig 

7, the effect of producing the membrane from different 

materials is investigated. Clearly nickel provides a good 

deal of rigidity, whilst also being a good conductor. Gold 

is the most flexible, suggesting that the applied force to 

give complete deflection is significantly lower. Potentially 

a gold membrane may be less durable than a nickel one, 

although whether this durability would be a problem for 

our devices in the long-term is uncertain. 

Deflection (microns) 

0.018 

0.016 

0.014 

0.012 

0.010 

0.008 

0.006 

0.004 

0.002 

Nickel nickel Gold Gold Tungsten Copper 

Material 

Figure 7: Variation in deflection under a fixed load for a 

typical 1000 micron long 5 micron thick 100 micron wide 

membrane with only variation of the material parameters.


0.005 

0.004 

0.003 

0.002 

0.001 

0.000 

0 10 20 30 40 50 

Beam Thickness (Microns) 


typical 1000 micron long 100 micron wide Nickel 

membrane with only variation of the membrane depth. 


0.035 

0.030 

0.025 

0.020 

0.015 

0.010 

0.005 

0.000 

0 50 100 150 200 250 300 350 

Beam Width (Microns) 


typical 1000 micron long 5 micron deep Nickel 

membrane with only variation of the membrane width. 


0.005 

0.004 

0.003 

0.002 

0.001 

0.000 

200 300 400 500 600 700 800 900 1000 1100 

Beam Length (microns) 


typical 5 micron deep, 100 micron wide Nickel 

membrane with only variation of the membrane length. 

Varying the membrane dimensions provides control of 

the deflection for a fixed applied pressure over the probe 

bump region. These results are given in Figs. 8-10, 

showing that increasing the thickness, increasing the 

width and decreasing the length of the membrane all 

make it stiffer. It is therefore possible to provide a degree 

of control to the deflection purely by varying the 

54 

membrane dimensions. However, it should be noted that 

there are also certain fabrication issues which would 

provide a limit on this, as discussed in 3. 

(a) 

(b) 

(c) 

Figure 11: Deflection under a fixed load for a typical 5 

micron thick membrane a) circular, b) ‘X’ shaped, c) ‘H’ 

shaped

Relative Deflection (Arb. U.) 

1.0 

0.8 

0.6 

0.4 

0.2 

0.0 

1 2 3 4 

H Circular X Beam 

Membrane Structure 

Figure 12: Deflection under a fixed load for a typical 5 

micron thick membrane of varying geometry 

Finally it is possible to simply examine geometrical 

variations of the membrane to determine the device 

behaviour under load. Three typical examples can be 

seen in Fig. 11. The circular device was chosen as it is 

an optimum shape for looking at solder bumps (Fig. 

11a). The ‘X’ and ‘H’ design were considered as they 

provide a design which is easier to remove residual 

material during the sacrificial layer removal stage, 

coupled with different control on the stress and 

deflection for a specific applied force (Fig. 11b-c 

respectively). If we then examine the deflection for a 

fixed force, we find that, as shown in figure 12, the 

deflection will vary depending on our membrane shape. 

3 EXPERIMENTAL DETAILS 

Experimental fabrication also places constraints on the 

device design. The column z plane dimensions are 

limited by the thickness of the photoresist. Beyond this 

the samples will start to overplate. The size of the upper 

membrane is primarily determined by device fabrication 

considerations. The larger the membrane area, the 

greater the associated problems with stiction are found 

at the device release stage. Also increasing the 

membrane area decreases the number of potential 

probes. However, too small a structure increases the 

stress in the membrane during deflection and limits the 

use of the self-limiting sensor. Our typical structures 

have a membrane dimension of 1000µm x 100µm x 5µm 

supported on 2µm high columns. The probe bump is 

100µm x 100µm, which is comparable in size to a typical 

IC solder bump or pad dimension. 

The individual probe devices were fabricated using a 

similar technique to that previously used to produce our 

capacitor structures [7]. Initially a seed layer of titanium 

(20 A), with a thin layer of nickel (500A), was deposited 

on a silicon substrate (Fig. 13A). This was then 

patterned using S1813 photoresist to define the regions 

where the anchor points and self-limiting sensor were to 

be fabricated. The sample was then placed in the gold 

(or nickel) electroplating bath using a current of 2mA, the 

columns were plated up to the top of the photoresist 

surface. A further layer of photoresist was added, with 

only the columns being plated this time (Fig. 13B). 

55 

Figure 13: Schematic diagram of the fabrication process 

In order to plate the membrane, a thin layer of gold 

(20nm) was sputter deposited onto the surface of the 

sample (Fig 13C). Above this a second layer of photo 

resist was spun and defined for the membrane region. 

Again the gold electroplating technique was used to fill 

this region to produce the membrane (Fig 13D). The 

third step was the fabrication of the probe bump region 

using the same technique as used to define the 

membrane (Fig 13E). 

Finally the sample then underwent the sacrificial removal 

stage to free the membrane device. The top layers of 

photoresist were removed using full field UV exposure 

and immersion of the sample in the developing solution. 

The thin layer of sputtered gold was removed using a 

fast wet etch in gold etching solution. The under-layer of 

photoresist was then stripped initially using acetone in 

an ultrasonic bath followed by iso-propanol. Then the 

sample was placed into the gold (or nickel) etch. This 

was then followed by a dip in sulphuric acid : hydrogen 

peroxide (1 :1) to remove any remaining photoresist and 

the titanium under-layer. The sample was then washed 

again in iso-propanol and release fully achieved by 

placing in an oven to give the final device (Fig 13F) 

Figure 14: SEM image of an individual electroplated 

membrane structure. Insert is an typical device with 20 

individual measurement pads.

We fabricated a range of the electroplated structures 

(example Fig. 14), varying in membrane dimensions 

from 100µm by 100µm to 1000µm by 1000µm. These 

showed good adhesion to the substrate via the support 

columns. In the unloaded state the samples show no 

electrical conduction between the probe bump and the 

self-limiting sensor suggesting the structure had been 

released. In order to verify that no waste material was 

left under the membrane an electrical probe was used to 

apply a force to the probe bump and gently bring it into 

contact with the self limiting sensor. This was cycled, 

showing that the devices can go through a 

load/unloaded cycle repeatedly without damage and that 

the self- limiting sensor can easily detect when the 

membrane is touching it. For an applied force giving full 

membrane deflection, the total membrane to self-limiting 

sensor resistance was 7 ?, with a low contact resistance 

of less than 2 ?. It was also noted that varying the 

sample dimensions changed the required force to touch 

the self-limiting sensor, and that the trends in this data 

were similar to those suggested by the modelling (Fig 

15). 

Relative Applied Force (Arb. U.) 

Contact Resistance (Ω) 

1.0 

0.8 

0.6 

0.4 

0.2 

0.0 

2.5 

2.0 

1.5 

1.0 

0.5 

0.25 0.50 0.75 1.00 

Relative Device Width (Arb. U.) 

Figure 15: Comparison of experimental data with that 

produced from modelling for variations in cantilever 

width. [squares correspond to experimental data, 

triangles to modelled data]. 

0.0 

0.000 0.001 0.002 0.003 0.004 0.005 

Applied Force (N) 

Figure 16: Measurement of the contact resistance as a 

function of applied force for typical membrane beam 

(1000 microns x 100 micron x 5 micron) 

The contact resistance was determined using fabricated 

aluminium pads to simulate IC pads. Increasing force 

was applied to the pads to determine the relationship 

between force and contact resistance. The probe bump 

is crucial for these measurements. Without this design 

56 

feature the membrane will not deflect giving a high 

contact resistance greater than 10 ?. Addition of the 

probe bump allows simple contact to the pad while the 

membrane deflects. The contact resistance varies as a 

function of applied force (Fig. 16), ultimately giving a 

minimum resistance of 0.5 ? for a force of 3.4mN which 

is comparable with results of others, e.g. [5]. 

Examination of the deflection of the membrane probed 

using an AFM tip showed that for a constant applied 

force the membrane deflects until it reaches the bump 

stop (Fig. 17). The internal structure in the signal 

suggests that the membrane is trying to resist the 

pressure applied by the probe. 

Delflection (V) 

0.20 

0.15 

0.10 

0.05 

0.00 

-0.05 

-0.10 

-0.15 

-0.20 

0 500 1000 1500 2000 2500 

Depth (microns) 

Figure 17: Measurement of the deflection of a typical 

membrane beam (1000 microns x 100 micron x 5 

micron), using an AFM probe to apply the force and 

monitoring the signal in scope mode. 

Using our simple fabrication technique, it was also 

possible to fabricate quickly a number of structures of 

different geometries. This allows us to tie in our work on 

the modelling side easily into the fabrication route 

allowing us to optimise a device. Figure 18 gives SEM 

images of comparable devices to those modelled earlier 

(Fig. 11).

(a) 

(b) 

(c) 

Figure 18: SEM image of a individual electroplated 

membrane structure, following the designs produced 

using CoventorWare, a) circular membrane, b) ‘H’ 

membrane, c) ‘X’ membrane 

A sample of experimental results for the fabricated 

devices can be seen in table 2. This clearly shows that 

the contact force depends on the sample geometry, and 

indicates the potential of integrating a design package 

into the fabrication route. However, the results from the 

modelled simulations are not in agreement with those 

found experimentally (Fig. 19). The general trend is 

similar, but the absolute values disagree. Also there is 

some discrepancy seen in the data for the circular 

membrane. This is likely to be caused by limitations in 

modelling real devices. 

57 

Sample Contact Force (mN) Contact 

Resistance (? ) 

‘H’ Sample 0.8 0.28 

‘X’ Sample 2.2 0.46 

Circular 

Sample 

2.2 0.51 

Beam Sample 3.4 0.50 

Table 2: Contact force and Resistance for various 

different device geometries. 

Contact Force (mN) 

6 

5 

4 

3 

2 

1 

0 

Circular X H Beam 

Sample Geometry 

Figure 19: Comparison of contact force for experimental 

data (triangles) and modelled data (squares) 

4 DISCUSSION 

Comparison of the modelling solutions provided using 

the CoventorWare software to our experimental results 

allow us to determine the use of the modelling in relation 

to our IC probe project. Clearly information can be 

provided on the deflection as a function of the design 

parameters. This is crucial in tuning the device to 

provide an opposing force to minimise the applied force 

needed to make a good contact with a low resistance. 

Also, it allows us to make an informed decision on the 

mechanical properties of different materials that can be 

used to fabricate the membrane. This can then be 

coupled with information on the electrical properties of 

the material to provide an appropriate solution to the 

problem. In our case materials in the range from nickel 

to gold were found to behave well both mechanically and 

electrically. However, the modelling data allows us to 

fine tune our devices, particularly in terms of the 

deflection. The self limiting sensor will provide a greater 

potential device lifetime when reaching the production 

stage. Also the nature of the software means that more 

exotic materials could be investigated for their potential 

in improving device stiffness prior to fabrication. 

The modelling also provides a good insight into the 

regions of stress. In some of the earlier fabricated probe 

devices, it became clear that the membranes were 

becoming damaged and torn from the support columns. 

This was believed to be due to the high stress regions 

near the column-membrane interface as found in the 

modelling solutions. Therefore by changing the 

dimensions of the beam, the damage was eliminated.

Finally, we are able to quickly implement a new design in 

the modelling package to determine if it has any 

potential benefits, and then fabricate quickly. This allows 

the design to be customised for any individual problem. 

In terms of negative aspects when producing our 

structures, while CoventorWare can be used to look at 

relative effects produced by design changes and to give 

the designer an idea of how these changes will affect a 

device performance, it is not yet at a stage where it can 

be used to produced a device of a specific performance. 

For example if we look at the applied force to give full 

deflection for a specific 1000 micron long, 5 micron thick 

beam, it can be found that the modelling suggests 6mN 

over a 100 micron square. However, an experimental 

value is of the order of 3mN. 

It is reasonable to expect problems in producing a fully 

workable model prior to any experimentation. It is not a 

simple process to accurately determine the material 

structure without some characterisation of the processes 

involved. Also possible imperfections in the device itself 

and the release process would provide further 

inconsistencies. It may be possible with further 

development to include these errors in the system to 

provide a more accurate model of the device process, 

but using current technology it is always going to be 

difficult to accurately produce a perfect estimate of how 

a device will perform in the real world. 

5 SUMMARY 

The work undertaken in this paper clearly shows the 

feasibility of fabricating a novel IC probe card structure. 

The use of the CoventorWare software allows 

optimisation of the design prior to fabrication, such that 

the process is significantly simplified. It also allows us to 

determine any weakness points in our sample, as well as 

giving us some idea how to control the deflection for a 

pre-defined applied pressure. Furthermore, the software 

allows us the potential to make quick design changes 

without the need for fabricating each individual design 

consideration. Comparisons between experimental and 

modelled results show similar trends in data, meaning it 

is feasible to input more control into the final device. 

However, it has to be recognised that the software does 

not easily take into account device limitations and 

imperfections. This means that there will be errors in 

absolute values. 

58 

Our electroplated design itself is particularly durable for 

industrial application, with multiple cycles showing no 

damage to the structure. The low contact resistance of 

our devices is consistent with that found in more 

complex structures. The structures have the potential for 

mass manufacturing in large arrays, with the additional 

benefit that they can be simply modified during the 

design phase to accommodate either contact pads or 

solder bumps. The relatively low production cost means 

that our devices have the potential to make a significant 

impact on the IC measurement market. 

6 ACKNOWLEDGMENTS 

This work is gratefully funded by One North-East via the 

County Durham sub-regional partnership, project 

SP/082. We would like to also thank the other members 

of the Microsystems Technology Group for input and 

assistance. This work has a patent (GB 0412728.8). 

6 REFERENCES 

[1] Park S. at al., 2002 A novel 3d process for singlecrystal 

silicon micro-probe structures, J. 

Micromech. Microeng. 12 No 5 pp 650-654, 

[2] Kataoka K. et al., 2003 Electroplating Ni micro 

cantilevers for low contact-force IC probing, Sens. 

Act. A: Phys., Vol. 103, issue 1-2, pp 116-121, 

[3] Cho Y. et al., 2004 Fabrication of Sharp Knife-edged 

micro probe card combined with shadow mask 

deposition, Sens. Act. A: Phys. In Press, 

[4] Kataoka K. et al., 2004 Multi-layer Electroplated 

Micro-spring Array for MEMS Probe Card 

MEMS2004 Proceedings, pp 733 -736, 

[5] Krüger C., Mokwa W., Schnakenberg U., 2004 NiW- 

Micro Springs For Chip Connection, MEMS2004 

Proceedings, pp 117 – 120, 

[6] CoventorWare Inc., 2001 Analyzer Ref. Guide, 

http://www.coventor.com 

[7] Gallant A.G., Wood D., 2004 The role of fabrication 

techniques on the performance of widely tunable 

micromachined capacitors, Sns. Act. A : Phys Vol 

110, issue 1-3,

CAD FRAMEWORK FOR THE DEVELOPMENT OF A THERMALLY 

ACTUATED MICRO-GRIPPER 

B. Solano*‡, D. Wood†, S. Rolt‡ and P.G. Maropoulos*‡ 

*Design and Manufacturing Research Group, School of Engineering 

University of Durham, Durham, UK, DH1 3LE 

† Microsystems Research Group, School of Engineering 

University of Durham, Durham, UK, DH1 3LE 

‡ IADET (Institute for Agility and Digital Enterprise technology) 

NETPARK Institute, Sedgefield, UK, TS21 3FB 

Abstract 

This paper describes the development of a simplified analytical and finite element model of a 

microelectromechanical (MEMS) device. The basic device used in this study is a thermally actuated microgripper 

that operates by differential thermal expansion caused by resistive (Joule) heating in its two 

constituent U-shaped micro actuators. The design and finite element analysis has been accomplished with 

the Conventorware TM Computer Aided Design (CAD) software specially developed for MEMS. This CAD 

software incorporates a material data base and a process description file that generates a 3D solid model of 

the device (virtual model) which provides a spatially detailed description of the system. Good correlation 

between the virtual and simplified thermo analytical model and experimental data validates the model and 

encourages its use for the evaluation of the system under different design configurations. 

Keywords: 

Computer aided design, CAD, MEMS, micro-gripper, thermal actuators 


Micro Electro Mechanical Systems (MEMS) is a relatively 

new way of making complex electromechanical systems 

using batch fabrication techniques adapted from the 

Integrated Circuit (IC) industry. MEMS are highly 

miniaturized devices that can include biological, optical, 

fluidic, magnetic, and other systems on a single silicon 

chip [1, 2]. Being inherently smaller, lighter, cheaper and 

faster than their macroscopic counterparts, MEMS 

devices can open new opportunities for product innovation 

in many different areas. Nevertheless, despite the clear 

potential of this new field the growth of the market has 

been slower than predicted in the past few years. There 

may be a number of reasons underlying this resistance to 

the emerging technology. These might include perceived 

risks, expensive fabrication, long development cycles and 

high costs associated with MEMS products. As in other 

mature disciplines, the incorporation of Computer Aided 

Design tools to the design process can help to reduce the 

long development times associated with MEMS. In 

contrast to microelectronics, a MEMS designer has to 

optimize three dimensional structures. The simulation 

approach, as opposed to the heuristic design approach 

that relies on the fabrication know-how of a few experts, is 

gaining wider acceptance amongst the MEMS community. 

These simulations not only enable the interpretation of 

limited experimental data but also enable the designer to 

explore the entire parameter space that influences the 

performance of a device. 

The modelling strategy [3] adopted in this paper is 

twofold. Firstly, to find a simplified analytical model to 

capture the essential device behaviour and to gain insight 

into its operation without employing excessive 

computational resources. Secondly, to use finite-element 

software to create a more rigorous and accurate 3D 

model (virtual prototype) of the system capable of 

providing an accurate and detailed description of the 

system with high spatial resolution. Good correlation 

between the virtual model and experimental data validates 

59 

the model, and encourages its use in the evaluation of the 

design-environment space: for example, using different 

structural materials, various surrounding media and 

diverse boundary conditions. 

2 THE MICROGRIPPER 

In this paper, we use this approach to produce a virtual 

prototype of a MEMS micro actuator. The basic device 

used for this study is an electro-thermally actuated microgripper 

[4].The principle of operation is the well 

established one-hot-arm thermal actuator. Figure 1 shows 

the basic scheme of a thermal U-shaped micro-actuator of 

this type. The actuation structure, composed of two 

adjacent micro beams, deflects in-plane at its tips by the 

asymmetrical thermal expansion of its constituent parts 

which have variable cross-sections and lengths. 

Thus, after the application of a current through the 

conducting layer, the thinner arm (hot arm) with a higher 

resistance will heat more than the wider arm (cold arm) 

and therefore will expand more. As both arms are 

connected at their free ends an in-plane deflection around 

Figure 1: (a) simplified one-dimensional coordinate 

system (b) in-plane deflection (c) Schematic of a 

traditional one-hot-arm actuator

the z axis will take place. The micro-gripper consists of 

two such micro actuators positioned face to face (see 

Figure 2). The actuating structure of the micro-gripper is 

composed of two layers: a structural dielectric layer of the 

polymer SU8 [5, 6] and a thin conducting layer of Cr/Au. 

The jaws are made of SU8 exclusively. The thermal 

expansion of the SU8 is achieved through the resistive 

heating of the gold layer. 

Given the biological end application of the micro-gripper, 

the maximum operating temperature will be limited to 

360K to avoid thermal degradation of the biological 

samples that have to be manipulated. Some physical 

properties of the polymer SU8 such as its biocompatibility, 

structural rigidity and large thermal expansion coefficient 

make it an ideal material for biological application inside 

physiological-aqueous media[5, 6]. The high coefficient of 

expansion permits low temperature difference activation 

and therefore low activation voltages that are necessary if 

electrolysis is to be avoided. Some experimental data has 

been reported in this direction for the manipulation of cells 

in-vitro. 

Figure 2: Schematic of the layers of the micro-gripper 

2.1 Modelling Background 

The type of thermal in-plane actuator (TIM) shown in 

Figure 1 has been extensively studied in the literature in 

its embedded actuation version, where the same material 

is used to produce the heating and the desired expansion. 

This in-plane actuator, usually fabricated from poly-silicon 

and separated from the substrate by a small air gap, has 

been demonstrated, and many analytical and finite 

element models (FEM) models [7] have been developed. 

Also topology optimization techniques have been applied 

to the design of this type of actuator[8, 9]. However the 

actuators composing our micro-gripper are not an 

embedded type TIM. Here the conducting layer is 

intended only to provide heating and not to produce any 

mechanical effects which, unlike the TIM, have to be 

avoided. Moreover, the overhanging configuration of the 

micro-gripper, together with the high thickness of the bilayer 

structure and the relatively small gap separating the 

arms, can affect the thermal behaviour of the system and 

therefore the steady-state temperature profile which 

causes the deflection. 

In this context, there exists the need to develop further 

such analytical and FEM models to characterise the 

micro-gripper more accurately. Here, we combine 

analytical and FEM approaches to fully characterise the 

deflection of the micro-actuator in steady state conditions. 

3 MODELLING THE MICROGRIPPER 

Imposing the correct boundary conditions in finite-element 

analysis is important because a change in the type of 

boundary condition (BC) can modify the simulated 

behaviour of the system. The deflection of the micro- 

gripper is controlled by the steady-state temperature 

profile developed along its constituent parts which in turn 

causes thermal expansion. As such, correct 

implementation of the thermal boundary conditions is 

crucial for the modelling of the micro-gripper. The first 

step towards establishing these conditions is to perform a 

detailed thermal analysis of the system. Moreover, a good 

understanding of the physics underlying the thermal 

behaviour of the device should provide more complete 

and accurate simulations. In addition, this understanding 

should promote a more efficient use of those simulations, 

and consequently a better performance of the overall 

modelling process 

3.1 1-D Electro thermal analysis 

Intrinsic to the functionality of the micro-gripper is its 

interaction with the environment and with the biological 

samples that have to be manipulated. An examination of 

this configuration indicates that, a priori, all three modes 

of heat transfer, namely conduction, convection, and 

radiation could take place. Conduction along the bi-layer 

beam structure to the anchors on the substrate, coupled 

diffusion-convection heat transfer to the surrounding 

fluidic media, radiation exchange with cooler 

surroundings, and also internal heat exchange between 

device elements at different temperatures (Figure 3 (b)). 

Since the characteristic dimensions are very much greater 

in the x direction, a one-dimensional model is adequate to 

describe the heat transfer mechanisms taking place in the 

system [10]. Thus, the heat transfer relations for each arm 

of the micro-gripper can be obtained by applying an 

energy balance to an infinitesimal volume element (w x z x 

∆x) of any of the arms (Figure 3 (a)) 

P 

60 

. . . 

g = q c + q dc + q r 

(1) 

where the term Pg is the electrical power generated in the 

volume element, qc is the heat rate lost by conduction, qdc 

is the heat rate lost by a coupled diffusion-conduction 

phenomenon and qr is the heat rate lost by radiation to the 

environment. At steady-state the heat generated within 

the infinitesimal volume will be entirely dissipated to 

maintain the steady-state temperature profile. 

Figure 3: (a) 1D infinitesimal volume element (b) Heat 

losses in steady-state (cross section view) 

The power Pg is generated by current flowing through the 

gold layer of the element which acts as a resistor and 

causes internal heat generation (Joule heating). The 

power dissipated by this element per unit volume per unit 

time can be calculated by using (2). In general, the 

resistivity of a conductor is temperature dependent. 

However, given the low range of operating temperatures 

(Tmax ≈ 360K) of the micro-gripper, ρ| ∆x can be 

considered,for practical purposes as a constant.

P g 

2 

I ρ 

= ∆x 

w z 

where I [A] is the current flowing through the conductor, 

and ρ| ∆x [Ω m] is the resistivity of the conductor element. 

The conduction term (qc) in (1), is defined as the addition 

of heat conducted through the volume element plus the 

heat conducted through the fluidic gap that separates the 

element and the opposed arm. The nature of this last term 

will be determined by the relative size of the gap between 

the arms and the wall height. Assuming the gap size to be 

smaller than the height of the wall (gap

Figure 4: 3D model of the thermal micro actuator generated by Coventorware TM 

Once the geometrical model has been built, a Manhattan 

or brick meshing with hexahedral second order elements 

is used in the model. The Manhattan Meshing has been 

chosen because it provides the most accurate results [11] 

for all the solvers within the software when the geometry 

of the 3D model is such that it only contains 90 degree 

angles as in the case of the micro-gripper under study. 

The elements were rectangular with variable dimension 

between 10x10x10 µm to a minimum size of 2x2x0.15 µm 

and 27-nodes. The actuator was modelled with 

approximately 6160 volume elements and 4528 surface 

elements. The air is modelled with roughly 85000 volume 

elements and 16000 surface elements. The size of the 

elements was chosen after a convergence analysis. 

Description Value (µm) 

Width hot arm (whot), SU8 7 

Width hot arm (whot), Gold 2 

Width cold arm (wcold), SU8 22 

Width cold arm (wcold), Gold 11 

Width flexure arm (wflexure), SU8 6 

Width flexure arm (wflexure), Gold 4 

Length hot arm (Lhot) 200 

Length cold arm (Lhot) 140 

Thickness (zSU8), SU8 20 

Thickness (zAu), Gold 0.3 

Gap 11 

Table 2(a): Dimensions for the micro gripper 

Description Units Value 

Young's modulus (E), SU8 GPa 4.4 

Young's modulus (E), Gold GPa 57 

Coefficient of thermal expansion (CTE), SU8 ppm/K 50 

Coefficient of thermal expansion (CTE), Gold ppm/K 14.1 

Thermal conductivity (κSU8), SU8 W/m K 0.2 

Thermal conductivity (κAu), Gold W/m K 297 

Thermal conductivity (κAir) W/m K 0.026 

Electrical Conductivity (σ), Gold (300K) S/m 4.40E+07 





Table 2(b): Thermo physical properties used in the 

simulation 

Fixing Boundary conditions 

Based on the results of the simplified thermal model, the 

boundary conditions to the behavioural analysis can be 

established 

• Electrical BC’s: 

Electric currents have been induced along the actuator 

using a constant-current boundary condition. The 

resistance of the gold circuit of the micro gripper is 19.6Ω . 

Thus an applied current of 10.2 mA to the contact pads of 

the gold circuit results in a generated input power of 

approximately 2mW. For the initial simulations, a 

constant-current boundary condition has been considered 

to be better suited than a constant-voltage boundary 

conditions, as it allows a comparison of both the 

simulation results and the experimental data, whilst 

excluding the influence of other parameters linked to the 

experimental set-up, e.g. the wiring. 

62

• Mechanical BC’s 

The anchor which provides mechanical support and 

connection to the substrate has been assumed to be fixed 

at all its sides. 

• Thermal BC’s 

The base of the anchor has been maintained at room 

temperature which implies that the anchor is thermally 

grounded by being in contact with a large thermal sink. 

Later in this paper we will evaluate the significance of this 

BC. 

Given the low value (∼ 10 -5 ) of the Grashoff number 

associated with a device of the size of the micro-gripper, it 

can be considered that the dominant phenomena at this 

scale is pure conduction as opposed to convection [12]. 

Having established this, a simple approach to model the 

device in air would be to place conducting air elements all 

around the micro-gripper [13] and to fix the outer faces of 

the air at ambient temperature. However, it has been 

numerically verified that to avoid influencing the simulation 

results it is necessary to simulate a cube of air of at least 

200x200x800 µm in volume. This is substantially larger 

than the scale of the gripper itself, and would require 

substantial computational resource to solve. Therefore, in 

order to avoid long computational times, another 

modelling approach is proposed to take into account the 

losses to air. Firstly, inclusion of a modest number of air 

elements in the gap between the hot and cold (flexure) 

arms will account for intra device heat exchange. 

Secondly, the inclusion of a general heat convection term 

or parameter for the exposed surfaces (see figure 5(b)), 

will describe convective/conductive heat loss to the 

ambient. To do that it is necessary to introduce in the 

software a heat transfer coefficient to air at the exposed 

surfaces. This coefficient can be obtained running one 

simulation with the whole volume of air and extracting the 

temperatures and heat losses in each surface. Thus using 

equation (6) h can be calculated. 

κ ∞ ∗ Heatloss 

at the surface 

h = 

T − T 

s 

∞ 

The average value obtained for the whole micro gripper is 

1070 Wm -2 K -1 . 

To finalize the implementation of the thermal boundary 

conditions, an effective heat transfer coefficient has been 

included at the end connection of the hot arm to account 

for the heat losses associated with the jaw of the microgripper. 

This coefficient has been analytically calculated 

and implemented in the calculations with a value of 

4450Wm -2 K -1 . 

• Material properties 

The electrical conductivity of the conducting layer is 

dependent on the temperature and the simulations uses a 

look-up table to access values. The corresponding values 

are shown in Table 2 (a). 

4 VALIDATION OF THE 3D MODEL 

The geometry of the micro-gripper used for the first 

simulation is identical to the geometry for which 

experimental data have been previously published [4].The 

most relevant dimensions are summarised in Table 2. 

The in-plane tip deflection of the thermal actuator was 

investigated by systematically applying increasing 

voltages to the contact pads at the anchors. For each of 

these simulations it is possible to extract simultaneously 

the deflection versus the applied power or the deflection 

versus the maximum or average temperature in the hot 

arm. For example, inspection of Figure 5 (a) reveals that 

for an applied current of 11.2 mA the deflection at its tips 

is 1.22µm. At the same time inspection of figure 5 (b) 

shows that the hot arm reaches a maximum temperature 

of 366K. 

Figure 5:3D Simulation results (a) Deflection with an applied current of 11.2mA 

Figure 5(b): Temperature distribution with an applied current of 11.2 mA (Left flexure and cold arm, right hot arm) 

63 

(6)

At this stage of the work, it is interesting to compare the 

simulation results with the available experimental data. 

Figure 6 shows the dependence of the deflection versus 

induced current in air. The data have been extracted from 

simulation and approximate experimental data. The net 

displacement in the simulation has been calculated at the 

end of the actuator (x=200µm) and at the middle of the 

hot arm. The y axis indicates the net displacement of the 

jaws of the micro gripper from its rest position. Depending 

on the initial separation of the two jaws, bodies ranging in 

size from two microns in diameter and upwards can be 

grasped. 

Displacement (um) 

12 

10 

8 

6 

4 

2 

0 

Approximate experimental data FEA Data in Air 

6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 

Currrent (mA) 

Figure 6: (a) Deflection versus induced current between 

the contact pads. 

There is a good agreement between the trends in the 

FEA results and experimental data. At low displacements 

the difference between the experimental results and 

simulation is bigger, but it is difficult to say whether this 

difference comes from the FEA model or from the 

measurements themselves. It will be possible to analyse 

this difference when our experimental results are 

available. 

5 MODEL-BASED SIMULATION PREDICTIONS 

The clear and substantial correlation between 

experimental and FE data encourage the use of the 

proposed model to study the influence of boundary 

conditions, material properties and geometry of the microgripper 

prior to its fabrication. 

5.1 Boundary conditions at the anchors 

The choice of the thermal BC at the anchors is important 

because it can modify considerably the overall heat 

losses in the device. Mankame et al [7] demonstrated that 

imposing a finite heat loss under the anchors of an 

embedded TIM via the introduction of an insulator (glass) 

can increase considerably the energy efficiency of the 

system. Theoretically, under the same applied input 

power, a higher temperature will be attained by the hot 

arm, increasing the overall temperature of the actuator. 

This may give a higher deflection for the whole system, 

although it is the temperature difference between the hot 

and cold arms that drives the deflection. Figure 7 depicts 

the anchor of the micro gripper and the associated 

thermal resistance with the introduction of a glass 

substrate. 

To avoid long computational times the glass substrate, 

with a thickness of 4 mm, is simulated by an effective 

64 

heat loss applied to the base anchor. This effective 

coefficient (h’) is given by 

Figure 7: (a) Anchor thermally grounded (b) Anchor 

thermally isolated 

R κ 

g = 1 / h' 

= z glass / glass 

(7) 

where κglass (0.96 W/m K) and zglass are the thermal 

conductivity and the thickness of the glass plate 

respectively. 

Using (7) the value obtained for h’ is calculated and is 

fixed in the simulations to a value of 245 Wm -2 K -1 . 

Temperature (K) 

353 

333 

313 

293 

Grounded Anchor Thermally Isolated Anchor 

0 50 100 150 200 250 300 350 400 

x position along the micro actuator (um) 

Figure 8 (a): Temperature profile (10.2mA) along the 

actuator under two different boundary conditions at the 

anchor (See Figure 1(a) for coordinate system). 

Displacement (um) 

FEA Data Anchor Grounded FEA Data Anchor Thermally Isolated 

12 

10 

8 

6 

4 

2 

0 

7.5 8.5 9.5 10.5 11.5 12.5 

Currrent (mA) 

Figure 8 (b): Deflection versus induced current between 

the contact pads from experimental data with two 

different BC’s at the anchor; anchor grounded and anchor 

Thermally isolated.

Figure 8(a) shows the resulting temperature profile along 

the micro gripper arms (see coordinate system in Figure 

1(a)) under two BC´s at the anchors representing a 

grounded anchor (Figure 7(a)) and a thermally isolated 

anchor (Figure 7(b)). Figure 8(b) shows the improved 

deflection of the micro gripper with an insulated anchor. 

As can be seen, for the same input power a 5% 

improvement in deflection is achieved. Bigger deflections 

could be obtained changing the thickness and the 

properties of the insulating layer. 

These results also indicate that the model with thermally 

isolated anchors can be chosen only when the associated 

higher operating temperatures are acceptable. 

5.2 Impact of the gap size 

In this section the influence of the size of the gap between 

the arms of the actuators is studied. The same electrical 

power of 2mW is maintained for three simulations with 3 

different gap sizes: 5.5µm (model 1), 11µm (model 2) and 

22µm (model 3). Figure 9 and 10 illustrate for each of the 

models the temperature profile developed along the 

actuators and the net deflection in the y direction. Table 3 

summarizes the results of the models. 

The temperature profiles presented in figure 9 indicate 

that the simplified electro thermal model proposed in 

section 3 correctly describes the thermal behaviour of the 

micro gripper. As predicted, the temperature difference 

between the hot and cold arm increase with the gap size. 

This fact validates the assumption of the simplified 

analytical model that considers a non-negligible amount 

of heat transfer through the air gap. This heat exchange 

together with the heat loss to the ambient dominates the 

heat transfer phenomena in the whole micro-gripper. 

Temperature (K) 

353 

343 

333 

323 

313 

303 

293 

Model 1 (gap= 5.5 microns) Model 3 (gap=22 microns) 

Model 2 (gap=11 microns) 

0 50 100 150 200 250 300 350 400 

x position along the micro actuator (um) 

Figure 9: Temperature profile along the arms of the micro 

actuators with a thermally grounded anchor and an 

applied current of 10.2mA 

Model 1 

(gap=5.5 µm) 

Model 2 

(gap=11 µm) 

Model 3 

(gap=22 µm) 

T maximum 342.61 347.31 352.08 

T average actuator 32.99 33.72 34.32 

T average hot arm (Th) 37.40 40.34 43.40 

T average cold arm (Tc) 31.13 29.81 28.10 

T average flexure 22.64 20.75 18.62 

Th-Tc 6.28 10.54 15.30 

Deflection maximum 1.18 1.22 1.05 

Temperatures relatives to the ambien temperature (293K) 

Table3: Temperature and deflection relevant data, same 

input power (2mW) for models 1, 2 and 3. 

65 

Figure 10: (a) Deflection model 1 (gap=5.5µm) with an 

input power of 2mW (b) Deflection model 2 (gap=11µm) 

with an input power of 2mW (c) Deflection model 3 (gap= 

22µm) 

The net deflection in the y axis for the model is shown in 

Figure 10 and reveals net deflections of 1.10µm, 1.13µm 

and 0.97µm for the models 1, 2 and 3 of respectively. 

These deflection results indicate that the impact of the 

end connection of the actuator is also important. Thus 

model 2 with a smaller gap (50% smaller than model 3) 

and a lower change in average change in temperature 

(32% lower) produce the highest deflection of the three 

models (16% higher) 

These results also reveal that for an increasing size gap, 

between 5.5 and 11 µm, the higher the difference in 

temperature between the two arms the higher the 

deflection. However this trend seems to be interrupted 

when the size of the gap approaches the size of 22 µm. 

Simulations run with other gap sizes indicate that a gap 

size of 11 µm is near the optimal value for an input of 

10.2mA

6 CONCLUSIONS AND FUTURE WORK 

Simplified electro thermal and FEA models for a SU8 

thermally actuated micro-gripper have been developed. In 

those models, the heat conducted between opposing 

arms of the actuator and the heat dissipated to the air, 

have been taken into account. The FEA model also takes 

into account the dependence of the conductivity with 

temperature. The good correlation of the FEA model with 

experimental data validates the model and encourages 

the use of it for the virtual testing of different conditions at 

the device anchors and under different geometries. It has 

been confirmed that imposing a finite heat loss at the 

base of the anchors increase the efficiency of the system. 

Equivalently it has been demonstrated that the size of the 

gap is of critical importance. Further work on a more 

detailed electro thermal model has shown promising 

results. We will shortly be validating those models against 

our experimental data. 

References 

[1] J. Judy, "Microelectromechanical systems: 

fabrication, design and applications," Smart 

materials and structures, vol. 10, pp. 1105-1134, 

2001. 

[2] J. M. Madou, Fundamentals of Microfabrication, 

2nd edition ed, 2002. 

[3] S. D. Senturia, Microsystem Design, 3rd Edition 

ed, 2001. 

[4] N. Chronis and L. P. Lee, "Polymer MEMSbased 

Microgripper for single cell manipulation," 

presented at Micro Electro Mechanical Systems, 

17th IEEE International Conference on. (MEMS), 

2004. 

[5] E. H. Conradie and D. F. Moore, "SU-8 thick 

photoresist processing as a functional material 

for MEMS applications," Journal of 

microelectromechanics and Microengineering, 

vol. 12, pp. 368-371, 2002. 

[6] "http://aveclafaux.freeservers.com/SU-8.html," 

1999. 

[7] N. D. Mankame and G. K. Ananthasuresh, 

"Comprehensive thermal modelling and 

characterisation of an electro-thermal-compliant 

microactuator," Journal of Micromechanics and 

Microengineering, vol. 11, pp. 452-462, 2001. 

[8] J. Jonsmann, O. Sigmund, and S. Bouwstra, 

"Compliant electrothermal microactuators," 

presented at IEEE Microelectromechanical 

Systems (MEMS), Orlando FL, 1999. 

[9] O. Sigmund, "Design of multiphysics actuators 

using topology optimization," Computational 

Methods Applied Mechanical Engineering, vol. 

190, pp. 6577-6604, 2001. 

[10] L. Lin and A. P. Pisano, "Bubble forming on a 

micro line heater," presented at MEMS, ASME 

Winter Annual meeting, Atlanta, 1991. 

[11] Coventor, "www.coventor.com," 2004. 

66 

[12] F. P. Incropera and D. P. DeWitt, Introduction to 

Heat Transfer, 3rd Edition ed, 1996. 

[13] A. A. Geisberger, N. Sarkar, M. Ellis, and G. D. 

Skidmore, "Electrothermal properties an 

modeling of Polysilicon Microthermal Actuators," 

Journal of Microelectromechanical Systems, vol. 

12, pp. 513-523, 2003.

The Applicability of CoventorWare to RF MEMS 

A.J. Gallant and D. Wood 

Microsystems Technology Group, School of Engineering 

University of Durham, Durham, England DH1 3LE 

Abstract 

RF MEMS devices are technologically well placed to enter the high volume market of wireless 

communications. However, simulation tools for the rapid virtual prototyping of RF MEMS are relatively 

immature. This paper examines the applicability of CoventorWare TM , commercial MEMS simulation 

software, to RF MEMS design. This is assessed through the case study of a widely micromachined 

tunable capacitor which has been developed at the University of Durham. The software was found to be 

effective at identifying approximate device operating regimes. However, the lack of accurate models for 

surface topology, reliability and RF simulation capabilities significantly impair its suitability for effective RF 

MEMS prototyping. 

Keywords: 

RF MEMS, varactor, microsystems, CoventorWare TM 


In recent years, primarily through the rapid uptake of 

mobile telephony, the wireless communications industry 

has been forced to adopt high volume production 

techniques. Continuous technological advances have led 

to exceptionally short product lifecycles which typically 

can be quantified in months. 

The functionality of the mobile telephone now extends far 

beyond basic wireless communications. The inclusion of 

on-board cameras and personal organisers reduces the 

space available for the radio frequency (RF) circuitry. In 

parallel, the consumer is demanding improved 

performance through better reception quality and battery 

lifetimes. 

Microelectronics achieves miniaturisation through single- 

chip solutions. These remove components from the 

printed circuit board and transfer functionality onto the 

microchip. However due to substrate and dielectric 

losses, many solid-state microelectronic solutions show 

poor RF performance. 

In recent years, microsystems technology has been used 

to miniaturise electromechanical RF devices. 

Micromachined RF switches, resonators, capacitors and 

inductors have all been demonstrated [1]. These are 

commonly referred to as RF MEMS (Radio Frequency 

MicroElectroMechanical Systems) and have shown 

improved performance over solid-state equivalents. 

RF MEMS devices share many fabrication processes 

with ICs and are therefore well suited for integration. This 

is extremely attractive to the wireless industry because it 

reduces the number of printed circuit board components, 

which in turn drives down manufacturing costs. 

This paper focuses on the electromechanical simulation 

of micromachined tunable capacitor structures using 

CoventorWare TM . This modelling package has been used 

extensively as part of an ongoing project to develop 

widely tunable capacitors. However, these devices 

successfully highlight many shortfalls associated with the 

use of CoventorWare TM for RF MEMS simulations. In 

particular, these include an inadequate representation of 

67 

surface topology and the lack of RF and reliability 

simulation tools. 

2 BACKGROUND 

Electrostatic tuning is preferred for RF MEMS structures 

because it enables high switching speeds whilst 

minimising power consumption. A key consideration with 

electrostatically tuned devices is an electromechanical 

instability known as pull-in [2]. This is where the 

electrostatic force of attraction cannot be countered by 

the mechanical restoring force. Electrical control is lost 

and the electrodes snap together. 

In RF MEMS switches, pull-in is advantageous because it 

is used to ensure a high contact force and therefore low 

contact resistance. In gap tuning capacitors, however, it 

can limit the tuning range. 

Gap tuning capacitors use the variation of the distance 

between two electrodes to alter the capacitance. An 

important advantage of the gap tuning capacitor is that it 

can be fabricated using relatively planar thin film 

technologies which are better suited to integration with 

existing ICs. 

Figure 1 shows a typical electrostatically biased parallel 

plate device configuration. As a voltage is applied 

between the movable driving electrode and the fixed 

ground electrode, electrostatic attraction reduces the 

gap. This increases the capacitance. 

Figure 1: A simple parallel plate micromachined capacitor 

arrangement

For stable tuning, the restoring spring force must counter 

the electrostatic force of attraction. Because of the pull-in 

phenomenon this can be shown to only occur for 

displacements of up to a third of the unbiased gap. 

Therefore, in an ideal, parallel plate, single gap capacitor 

the maximum tuning ratio is 1.5:1. In reality, however, 

the ideal arrangement may not be a good approximation. 

Deviation from parallel plate scenario can have 

significant effects on pull-in. CoventorWare TM readily 

enables the simulation of curved electrodes. The 

following section demonstrates its capabilities through 

the simulation of three capacitor designs. 

3 SIMULATION 

3.1 The single-gap tunable capacitor 

Figure 2 shows the dimensions of a single-gap parallel 

plate micromachined tunable capacitor fabricated using 

two gold electrodes. The movable upper electrode (1µm 

thick) is suspended 1µm above a lower electrode. The 

lower electrode is fixed to a substrate, which is not 

shown. In order to retain the generality of the results, the 

substrate is excluded from the simulations. The material 

properties used in the model are a Young’s modulus of 

57 GPa, a Poisson’s ratio of 0.35, a density of 19300 

kgm -3 and zero residual stress. 

In any microelectromechanical simulation, at least part of 

the simulated structure needs to be tethered to a fixed 

point. This forms a boundary condition for the solver. 

CoventorWare TM enables particular elements or groups 

of elements to be fixed in space. The upper electrode in 

this example has its four edges clamped. 

If a voltage is applied across the two electrodes, the 

upper electrode will be electrostatically attracted to the 

lower electrode. Due to the clamping, the upper electrode 

cannot move parallel to the lower one. Consequently, the 

upper electrode has to deform in response to the 

electrostatic force. 

1µ m 

1µ m 

Lower electrode 

Upper electrode 

500µ m 

Clam e 

ped 

edg 

edge ped 

Clam 500µ m 

1µ m 

Figure 2: A single-gap, two electrode tunable capacitor 

Figure 3 shows the displacement of the upper electrode 

subjected to a range of biasing voltages. This has been 

simulated using the CoSolveEM solver. The electrode 

can be seen to be deforming most at the centre, i.e. the 

point furthest from the clamping points. The edge 

clamping provides the restoring force for the upper 

electrode. 

68 

Figure 3: Displacement of the upper electrode in the 

single-gap device subject to a range of biasing voltages 

(displacement units are µm) 

Above 6V, CoventorWare TM predicts pull-in. Figure 4 

shows the tuning characteristic for this capacitor. A 

tuning ratio of only 1.15:1 is achieved prior to pull-in. This 

is less than the 1.5:1 which could be achieved in a purely 

parallel plate situation. The reason behind this is that 

towards the edges of the upper electrode, the 

displacement during biasing tends to zero. The edges of 

the upper electrode are contributing the least to the 

increase in capacitance. In the parallel plate scenario, the 

edges are free to contribute to the capacitance change. 

It should be noted though that the maximum deflection of 

the upper electrode, prior to pull-in, is 0.43µm. This 

exceeds the predictions of the parallel plate situation, 

which would allow 0.33µm deflection prior to pull-in. 

However this is a local maximum and at the clamped 

edges the displacement is zero. 

Capacitance [pF] 

2.60 

2.55 

2.50 

2.45 

2.40 

2.35 

2.30 

2.25 

2.20 

0 1 2 3 

Biasing voltage [V] 

4 5 6 

Figure 4: The capacitance tuning characteristic of single 

gap device 

3.2 Leverage based pull-in avoidance 

Figure 5 shows a view of an alternative tunable capacitor 

arrangement. It is a beam with only one end clamped, 

therefore forming a cantilever. There are five lower 

electrodes, labelled A to E. Electrode A and the 

cantilever form the capacitor electrodes whereas 

electrodes B to E are used for DC biasing.

The upper electrode is 500µm by 200µm. The lower 

electrodes are 50µm by 200µm with a 50µm spacing. 

The metallisation thicknesses and unbiased air gaps are 

both set to 1µm. 

Figure 5: A cantilever based tunable capacitor 

The principle of operation is that of a lever, where a 

deflection near to the clamping end corresponds to a 

larger deflection at the free end of the upper electrode. In 

this design, the biasing and capacitive electrodes have 

been separated. This enables the capacitive electrode to 

be positioned at the free end of the upper electrode. The 

biasing electrode can then be placed nearer to the 

clamped end. There will reach a point when the biasing 

electrode will be able to displace the upper electrode to 

contact the lower capacitive electrode whilst avoiding 

pull-in. 

Figure 6 shows the tuning characteristics of the device. 

When only electrode B is biased, the tuning ratio is 

1.39:1 prior to pull-in. 

When only electrode C, which is nearer to the clamping 

point than B, is biased the maximum tuning ratio prior to 

pull-in is 1.72:1. This is an improvement, although the 

peak tuning voltage has increased from 1V to 2V. 


0.35 

0.30 

0.25 

0.20 

0.15 

0.10 

Electrode B 

Electrode C 

Electrode D 

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 

Biasing voltage [V} 

Figure 6: The capacitance tuning characteristics of the 

cantilever device with electrodes B, C and D used 

individually for biasing 

Using only electrode D for biasing, pull-in occurs after the 

capacitive electrodes have made contact and has 

therefore been eradicated. This, in principle, enables any 

capacitive gap to be set which leads to an infinite tuning 

ratio. 

Figure 7 shows the displacement of the device subject to 

a range of electrode D biasing voltages. The simulations 

69 

have been performed using only a single electrode for 

actuation. However, applying a voltage to each electrode 

in turn could create discrete capacitances. This would 

form the basis of a digital variable capacitor. 

Figure 7: Displacement of the cantilever device subject to 

a range of biasing voltages on electrode D (displacement 

units are µm) 

The cost of using this type of leverage bending is twofold. 

First, the force required to deflect the beam increases 

with proximity to the clamping point. This can lead to 

higher control voltages. Second, the separation of the 

capacitive and biasing electrodes consumes more 

substrate space leading to larger devices. 

3.3 Two-gap based pull-in avoidance 

An alternative approach to pull-in avoidance is shown in 

figure 8. As in the previous section, this uses separate 

capacitance and biasing electrodes. However, it also 

uses a two-gap arrangement. 

A DC bias is applied between the upper electrode A and 

the lower electrodes, B. The upper electrode 

electrostatically deforms to reduce the gap between it 

and electrodes B. Electrodes A and C form the capacitive 

structure. As the upper electrode deforms, so the gap 

and hence the capacitance varies. If near contact can be 

achieved between the capacitive electrodes before pull-in 

occurs then very small gaps, and hence high 

capacitances, can be realised. 

Figure 8: The two-gap electrostatically tuned 

micromachined capacitor 

The horizontal dimensions for the simulations are shown 

in figure 8. For the purposes of the simulations, electrode 

A is clamped at each end. The upper electrode is 3µm 

thick. These results are based on an all nickel structure. 

In these simulations, nickel has a Young’s modulus of 

220.5 GPa, a Poisson’s ratio of 0.3 and a density of 8910 

kgm -3 . Figure 9 shows the tuning characteristics for a

ange of capacitive gap dimensions. In each case the 

biasing gap is set to three times the capacitive gap. 


5 g s =0.1µm 

4 

3 

2 

1 

g s =0.25µm 

g s =0.5µm 

g s =0.75µm 

g s =1µm 

0 

0 20 40 60 80 100 120 140 160 

Bias voltage [V] 

Figure 9: The simulated C-V characteristics of a two-gap 

capacitor for a range of capacitive electrode spacings, gs 

With this configuration, the two-gap approach is shown to 

have eliminated the effect of pull-in. However, in order to 

achieve the high tuning ratio, the electrodes need to be 

brought into very close contact. Furthermore, this graph 

illustrates that, for this configuration, unbiased submicron 

air gaps are required to achieve low voltage 

operation. 

4 FABRICATION 

For IC manufacture, many of the fabrication routes are 

standardised. For MEMS manufacture, this is not the 

case. Many competing process flows exist and currently, 

they tend to be device specific. 

Consequently, CoventorWare TM can readily identify key 

parameters which influence device operation, such as 

material properties, dimensions and electrode gaps. 

However, it is unable to suggest suitable fabrication 

routes. This lack of standardisation is an industry-wide 

problem which needs to be addressed for the successful 

virtual prototyping of general MEMS devices. 

For the MEMS process engineer, a range of mechanical 

considerations need to be considered in addition to the 

simulations. These are illustrated for the two-gap 

structure in figure 10. 

Figure 10: General fabrication considerations for the 

fabrication of the two-gap capacitor 

70 

A primary consideration is the self-compatibility of the 

flow. This is from both a chemical etch and mechanical 

perspective. For example, the structural metals need to 

be inert to the sacrificial layer etchant. From a 

mechanical perspective, the total stress on the wafer at 

any given point during process must not cause structural 

cracking or delamination. The micromachined structure 

needs to be well attached to the substrate for effective 

operation. 

Figure 11 shows the fabrication route for the two-gap 

tunable capacitor. A titanium sacrificial layer is used with 

either electroplated nickel or gold forming the capacitor. 

Specific processing details can be found in the literature 

[3]. 

Figure 11: The fabrication of the two-gap capacitor 

The titanium defines the unbiased electrode gaps. These 

are set to 0.3µm and 0.1µm for the bias and capacitive 

gaps respectively. In practice, thin film stress in the upper 

electrode can cause deviation from these gaps in 

released devices. 

Figure 12 shows a C-V tuning characteristic from a 

fabricated nickel capacitor. This has a tuning ratio of 

5.1:1 from an unbiased capacitance of 0.7pF. These are 

compared to simulated results for both an unstressed 

device and one with an additional 0.1µm of curvature 

representing a stressed state. 


3.6 

3.2 

2.8 

2.4 

2.0 

1.6 

1.2 

0.8 

Simulated 0.1µm gap 

Simulated 0.2µm curved gap 

Measured results 

0.4 

0 2 4 6 8 10 12 

Bias voltage [V] 

Figure 12: The C-V characteristics of a nickel two-gap 

capacitor compared to simulated results

The simulations are in the correct voltage regime. 

However, the prediction of the tuning characteristic is not 

perfect. 

CoventorWare TM has been successful in identifying the 

pull-in avoidance associated with the two-gap structure. 

Gold devices using this process flow have been 

demonstrated with tuning ratios up to 7.3:1 [3]. However, 

iesults are limited by the accuracy of the model. The 

remainder of the paper identifies key areas which are not 

effectively simulated by CoventorWare TM . 

5 LIMITATIONS OF COVENTORWARE 

5.1 Topological accuracy 

The two-gap capacitor forces MEMS simulation to its 

limits. First, large tuning ranges are only achievable if the 

electrodes can be brought into near contact. Second, the 

sub-micron gaps required to achieve low actuation 

voltage devices produce very high aspect ratio devices. 

For most finite element analysis, the surface topology is 

insignificant when compared to the dimensions of the 

simulated device. However, for the widely tunable 

capacitor, the surface topology of the electrodes 

becomes a significant consideration. 

Figure 13 shows an electron micrograph of an anchor 

region of a released nickel tunable capacitor. The 

surface roughness associated with the electroplated 

metal is visually apparent in both the anchor and upper 

electrode regions. It is therefore, anticipated that this 

topology would be evident in the critical capacitive region. 

Upper electrode 

Anchor 

Figure 13: A SEM image of the anchor region of a 

fabricated capacitor 

In order to quantify the topological features of the 

electrodes, an atomic force microscope (AFM) was used. 

Owing to scan area limitations, it was not feasible to track 

identical feature areas through the structure. However, 

general trends in surface roughness and topology 

characteristics could be studied. 

Surface roughness analysis in-situ of the underside of the 

upper electrode is not currently possible. To obtain this 

data the upper electrodes were lifted off the substrate 

using adhesive tape, and then examined with the AFM 

71 

[4]. Figure 14 shows the underside of a nickel upper 

electrode in the region above the bare silicon. In this 

case the mean surface roughness is 73Å. This 

roughness is primarily associated with the 3000Å of 

evaporated titanium. 

Figure 14: The surface topology of the underside of the 

nickel upper electrode in the region above bare silicon 

This demonstrates that the finite surface roughness of 

the titanium transfers through to the underside of the 

upper electrode. Figure 15 shows the surface topology of 

an electroplated nickel central electrode. A smooth metal 

grain is evident from the AFM scan. The mean surface 

roughness is 230Å. 

Figure 15: An AFM scan of the electroplated nickel 

central electrode 

Figure 16: An AFM scan of the underside of the upper 

electrode over the central electrode region

Figure 16 shows an AFM scan of the underside of the 

upper electrode in the region above the central electrode. 

This is essentially an inverted version of figure 15 with 

the lower electrode grain still visible. However, in addition 

the localised roughness associated with the evaporated 

titanium is also present, superimposed onto the grain 

pattern of the electroplated metal. The mean surface 

roughness in this region is 170Å. 

Figure 17: An illustration of the topology of the capacitive 

electrodes 

Figure 17 illustrates the model proposed for surface 

interaction in the electroplated capacitive structures. It 

shows the localised topology of the sacrificial titanium 

transferring to the underside of the upper electrode. 

Second, it shows a reduction in the overall surface 

roughness observed in the upper electrode. 

Stiction failure is where two surfaces irreversibly adhere 

together. This can occur during device release due to the 

capillary forces associated with drying liquids. 

Alternatively, it can occur during device operation due to 

the van der Waal forces which become significant with 

sub-micron gaps. 

Stiction is related to contact area [5]. In these devices, 

complete meshing cannot occur, which effectively 

reduces the absolute contact area and consequently the 

likelihood of stiction failure. 

However, the dominant roughness associated with the 

electroplated metal is significant from a capacitive 

perspective. The two-gap capacitor operates in a near, 

rather than an absolute, contact state. This roughness 

increases the capacitive area when compared to smooth 

electrodes. Therefore, at near contact, when compared to 

simulations, higher capacitances should be achieved. 

This is confirmed by the results in figure 11. 

These topological considerations are dependent upon the 

process flows and materials used for fabrication. 

Therefore, specific models need to be developed which 

address these issues. This is beyond the scope of 

conventional general purpose MEMS simulation tools. 

5.2 RF simulation 

The electromechanical simulations of the tunable 

capacitor provide an indication of tunability. However, 

CoventorWare TM does not address the RF performance. 

A design compromise exists between RF and mechanical 

requirements. 

72 

In addition to the tuning range, the quality factor, Q, is an 

important figure of merit for micromachined capacitors. A 

low Q is indicative of a high loss device. High Q devices 

exhibit sharp peaks in voltage controlled oscillators 

(VCOs) and tunable filters. This enables the RF spectrum 

to be used more efficiently and power consumption to be 

lower. 

The process flow presented here produces relatively low 

Q devices because they are fabricated on low resistivity 

silicon. A modified process flow which enables thick gold 

interconnect (6.2µm) in conjunction with the use of a 

glass substrate has produced a high measured Q of 243 

at 0.9 GHz for a 0.26pF capacitor [6]. 

The inclusion of RF simulation in a MEMS FEA package 

such as CoventorWare TM would make it more applicable 

to the design of the capacitor and RF MEMS in general. 

5.3 Reliability 

The implementation of an emerging technology such a 

RF MEMS into a volume market such a mobile telephone 

handsets requires excellent device reliability. 

CoventorWare TM can be used to identify regions of high 

stress which are likely to fail. However, the long-term 

issues such as metallic creep, fatigue and wear remain 

unsimulated. Furthermore, the effect of the environment 

on device performance is not well understood. 

These issues need to be addressed not only for RF 

micromachined devices but for the prototyping of MEMS 

in general. 

6 SUMMARY 

This paper has demonstrated the use of CoventorWare TM 

to simulate micromachined tunable capacitor structures. 

Its key functionality is found in electromechanical 

simulations of non-parallel plate devices. Two-gap 

structures have been shown to avoid pull-in associated 

tuning limitations in both simulated and fabricated 

devices. Tuning ratios up to 7.3:1 have been 

demonstrated for gold devices. 

The fit of the simulated and measured tuning curves is in 

the correct voltage regime, but not an absolute. 

Topological studies have demonstrated that the 

approximations used in standard FEA simulations are 

oversimplified for these two-gap structures which operate 

in a near-contact state. The surface roughness 

associated with electroplated metals and the titanium 

sacrificial layer is a significant contributor to device 

performance. 

At the moment fabrication flows are independent of the 

virtual design process. These need to be more closely 

integrated. This in turn will enable a realistic 

representation of topology.

Finally, a compromise exists between a good 

microelectromechanical and RF design. These trade offs 

need to be modelled by packages such as 

CoventorWare TM to really benefit the virtual prototyping of 

RF MEMS. This needs to be combined with 

improvements to the reliability simulation capabilities. 

7 ACKNOWLEDGEMENTS 

The authors would like to thank Filtronic Compound 

Semiconductors Ltd and EPSRC for funding this work. It 

is also funded by the regional development agency One 

North East as part of the University Innovation Centre in 

Nanotechnology. 

8 REFERENCES 

[1] J. J. Yao, 2000, RF MEMS from a device perspective, 

J. Micromech. Microeng., 10, pp R9-R38. 

73 

[2] A. Dec and K. Suyama, 1997, Micromachined 

varactor with wide tuning range, Elec. Lett., 33, pp 922- 

924. 

[3] A.J. Gallant and D. Wood, 2004, The role of 

fabrication techniques on the performance of widely 

tunable micromachined capacitors, Sens. Act. A : Phys., 

110, no 1-3, pp 423-431. 

[4] A.J. Gallant and D. Wood, 2004, Sacrificial layers for 

widely tunable capacitors, IEE Proc. Sci., Meas. & Tech., 

151, no 2, pp 104-109. 

[5] N. Tas, T. Sonnenberg, H. Jansen, R. Legtenberg, 

and M. Elwenspoek, “Stiction in surface micromachining”, 

Journal of Micromechanics and Microengineering, 6, pp. 

385-397, 1996. 

[6] A.J. Gallant and D. Wood, 2004, Widely tunable 

capacitors: The challenges of integration, Proc. IMAPS 

MicroTech 2004, Cambridge, UK, March.

PART 4 

PROCESS AUTOMATION 

74

Concurrent Electrical Engineering Methods 

N.S. Zughaid, P.D Hackney 

CIM-TEAM UK LTD, Poynton, Stockport, England, SK12 1QS. 

Abstract 

This paper looks at the benefits and practical methods of accomplishing concurrent electrical engineering design tasks, using 

today's design tools such as E³.series Enterprise, from CIM-TEAM GmbH. 

Concurrent engineering has become a modern day concept used liberally to convey the ability for multiple stages of a 

products design phases to be carried out simultaneously. A number of tools and forums use collaboration, where engineers, 

sometimes, at differing sites come together and collaborate, usually in WEB space to bring together different aspects of the 

design process in a single location. Here teams from concept to launch can work together in a dedicated space. Once the 

individual processes are removed from the collaboration area, they become remote and isolated. CIM-TEAM’s E³.series 

offers a true multi-user environment. 

No other ECAD tool on the market can be said to be truly multi-user, most offer a form of concurrency through batch updates 

or remote procedures. E³.series with its SQL database engine offers engineers true concurrency where Concept Engineers 

and Systems Engineers and Design Engineers can all work side by side in the same E³ Project. Concept Engineers can be 

laying down foundations, building blocks whilst System engineers begin to fill in the detail and Design Engineers concentrate 

on the detailed design. At any stage in the process all Engineers have the power to influence the design, and the other 

groups can immediately see any changes or innovations. 

The flexibility and power, true, concurrency has to offer today’s Engineers, through E³.series, has yet to be fully explored. 

However, through writing and delivering this paper it is hoped that a number of the possibilities can be discovered and 

presented in a practical manner. 

Keywords: 

Computer Aided Engineering, ECAD, Multi-user, E3.Series, CIM-Team 

1. INTRODUCTION 

1.1 ECAD DEVELOPMENT PHASE 

It is safe to say that there has been an explosion of ideas 

and concepts for CAD systems in the past 20 years, drawing 

boards, once the proud upstanding pillars of a drawing office 

have been thrown out or burned with the odd lucky ones 

residing in a dusty cupboard, brought out to show the new 

recruits how it used to be done. Electrical Cad has been 

around for the majority of the past 20 years, it is only in the 

last 10 years that a strong position of its own has been 

established. Mechanical CAD on the other hand has been 

dominant in the drawing office and still today many Electrical 

Designers are forced to use the mechanical package to carry 

out complex electrical designs, much more suited to a 

bespoke ECAD Package. The reason for this is quite simple 

in most organization Mechanical Engineers out number 

Electrical Engineers by about four to one therefore the 

requirements of the masses will be catered for. 

In the past five years software developers eager to push 

back and expose new boundaries have developed 

collaboration tools, these tools panacea is the involvement of 

supplier and customers throughout the lifecycle of a products 

development phase, all pulling together to reduce time to 

market and cut down errors and ultimately reduce costs. 

Another obvious benefit in the world of multi-nationals is the 

ability to have departments from different countries working 

together in the same environment to speed up development 

time and reduce travel requirements. 

These developments eventually filter down to the Electrical 

CAD market and we are now beginning to see dynamic tools 

allowing for concurrent Electrical Engineering. In the past 

75 

large electrical schematics have been broken down in to 

smaller sections and engineers and drafts-people have 

been able to work on separate jobs. The linkage between 

the schematics generally, if at all, being carried out as a 

post-process operation. The new truly dynamic breed of 

concurrent software tools available now offers a true, realtime, 

multi-user environment. 

1.2 ECAD BACKGROUND 

An Electrical Project consists of an entire set of schematic 

sheets. The design on these sheets show the electrical 

connectivity of all the components which make up the 

products design, be that a Power Station, a Piece of 

machinery, or a vehicle harness. The components 

contained within the Project are generally split into 

several Electrical symbols, which can be shown in an 

electrically intelligent manner on separate sheets 

throughout the Project. On an intelligent ECAD system 

these symbols are all linked together through a series of 

dynamic cross-references all updating to each other. Also 

shown within a Project are all the Devices used to 

produce the complete product these are displayed using a 

device tree. E3.Series from CIM-Team is a multi-module 

tool offering a suite of tools for the aid in ECAD and 

Pneumatic/Hydraulic designs. Figure 1, shows the 

modules currently available from the product.

Figure 1: E3.Series Product Suite 

2. MULTIUSER ENVIRONMENT THROUGH A CENTRAL 

SERVER PROCESS 

Multi-user functionality can be achieved using two different 

methods. Either with the CAD program directly working with 

one database which saves all changes and thus always 

contains the project’s current state, or this can be achieved 

using a central server process, connecting the users working 

on the client computers and synchronizing all information 

between the connected clients. The first variant has the 

disadvantage that the user program has to read all data 

directly from or write all data directly to the database. This 

can lead to speed issues with the large amount of data 

contained in some projects. The second method using a 

special server process, shown in Figure 2: also writes data to 

the database but systematically distributes the data to the 

clients. Tests concluded that controlling the data via a central 

server is the better solution for working in multi-user mode. 

This is due to the reduction in data transfer between the 

server and the clients, minimizing the network load. 

One important request from the user is that his/her operation 

mode does not differ, whether they are working on a project 

exclusively (single-user) or together with other colleagues, 

collectively working on a multi-user project. Both operation 

modes must be possible with the same underlying system, 

any differences must be transparent to the user. What is 

more, the transfer of projects between the different working 

methods has to be possible as the user must have the 

possibility to take out complete projects or parts of a project 

in order to work on them remotely, and then transfer it back 

to the server. And in doing so, the user wants to work on his 

single and multi-user projects using the same program, the 

same functionality and the same user interface. 

Figure 2: shows the multi-user topology. Within the SQL 

database are the Dictionary Tables, which contain the 

system information, which tracks and controls all the 

projects, the users and the Host services. From the diagram 

we can see that multiple Hosts can be setup, the various 

clients connect to the services setup by the hosts. When new 

multi-user projects are Projects are created, a new Database 

is created in the SQL environment. Each Project then has it’s 

own Project Service which manages the transactions from 

the clients. 

76 

Figure 2: Central Server Process. 

3. OPERATIONAL OVERVIEW 

A typical operational procedure is as follows: After 

opening a Project the user can immediately see of all 

objects used in that Project, e.g. sheets and devices, and 

at this level they can modify the devices and the sheet 

header information. All information which does not require 

graphic access to the sheets can be modified. This 

corresponds to the table-like modification of information in 

a database. In addition to this, reports concerning the 

overall project can be created anytime without having to 

check-in all sheets from the server first. Next the user can 

check-out the sheets to be graphically edited from the 

server and create new sheets, Figure 3: shows the tree 

structure showing the sheets being used and the ones still 

available for check-out. The users’ active sheets, i.e. 

sheets checked-out by this user, are locked for other 

users in the same project and cannot be edited. However, 

this only applies to the sheet’s ‘graphic representation’. 

The other users can still work with the (shared data) 

objects such as sheets or devices. User ‘A’, for example, 

can rename a device designation at any time, even if a 

symbol of this device is placed on a sheet being edited by 

user ‘B’. The new device designation will immediately be 

displayed on user ‘B’s opened sheet and User ‘B’ can 

continue working with this new information. New devices 

added to the project can immediately be used by all users 

working within the project. 

If the user has finished modifying the sheets, he can 

check-in these sheets to the server and thus unlock them 

allowing for ‘graphic modification’ by other users. The 

‘logic modifications’ are transferred immediately to the 

server.

Figure 3: Multi-User Sheet Tree 

4. POSSIBLE SCENARIOS 

This kind of simultaneous working is not only possible for 

large projects whose circuit diagrams are edited by several 

colleagues simultaneously. With smaller projects, several 

colleagues could be involved in editing different sections of 

the project. Some examples of this are the panel layout and 

panel wiring, the documentation for the cabling and 

pneumatics/hydraulics. This is especially true if the individual 

colleagues are in different design departments and are not 

generally working in close physical proximity. A change could 

be made to the schematic which affects the panel layout. In 

the multi-user environment this change can be picked up 

immediately by the panel engineer and that change acted 

upon if necessary. Prior to multi-user this change could have 

at the worst been missed, but at the best would have to be 

manually modified to bring it back into line with the 

schematic, taking time and costing money. Of course this is 

also true in the case of Pneumatic/Hydraulic schematics; 

changes to a coil’s name, function or even placement in the 

Electrical schematic will affect the valve in the 

Pneumatic/Hydraulic section. Communication between 

colleagues in the same department is not always as good as 

it could be, and this is especially true between separate 

departments. So, for this information to be updated 

dynamically, this is an obvious benefit. This is all made 

possible by having alternate views of objects, changes to any 

view of an object are reflected in all other views including the 

original view, combining this functionality with the multi-user 

environment provides a fully functional concurrent work place 

77 

where shared data really can be dynamic. Having a 

standard platform and being able to share projects and 

objects with colleagues who previously would have 

worked remotely will reduce development time and allow 

for checking to be carried out during the design phase. All 

this is possible without any subsequent effort or 

adjustment. 

4.1 ELECTRO-PNEUMATIC 

Figure 4: shows a pneumatic sheet within a Project, the 

electrical section of the valves shown in this schematic 

are located on other sheets. Keeping the information in a 

readable format, with separate Electrical and 

Hydraulic/Pneumatic sections but intelligently linking the 

associated symbols. 

Figure 4: Pneumatic/Hydraulic schematic. 

Figure 5: shows the combining of a pneumatic valve and 

the electrical solenoid, each symbol has the same Device 

ID –Y1 and a reference is shown between the two 

symbols, all stored under the one part number. 

Figure 5: Electrical/Pneumatic combination

4.2 SCHEMATIC-PANEL 

Figures 5: shows a fully wired panel layout in E3.Series, the 

routing of the wires can be based on the electrical schematic. 

Individuals can be placing the panel elements whilst others 

are producing the schematic section of the Project, both 

working together to speed up production. 

Figure 5: E3.Series Panel Layout. 

4.3 WORKING WITH ALTERNATE VIEWS 

Figure 6a: Original View of Cable –W2 

Figures 6a and 6b show how differing views of elements 

within an ECAD system can look. In Figure 6a we can see 

how the schematic shows the cable running between 

connectors –XS1 and –XP1, as far as the design engineer is 

concerned, that is enough information. However, the 

production engineer would like to look more closely at the 

make-up of cable –W2. So, by creating alternate views of 

cable –W2, shown in figure 6b: we see a documentation 

style of layout preferred by production. As the views are 

dynamically linked, any changes in one view reflect in the 

other and so once again combining this functionality with a 

multi-user technology allows Design and Production 

Engineers to share and modify the same Project data. 

During the design phase, production can begin the creation 

of the documentation, using their production skills, and may 

spot errors not usually seen by Designers, those errors can 

be rectified and the changes filter through to the Design 

section of the schematic. This form of Quality assurance is 

similar to techniques used in the production line where the 

78 

first job of the assembly line Engineer is to check the work 

carried out by the previous station. Adopting this 

methodology, could save time and money by spotting 

errors earlier in the design process, prior to assembly. 

Figure 6b: Alternate Views of Cable –W2 

Figure 7: shows the finished article, we can see from the 

picture how the product has a mixture of 

Pneumatic/Hydraulic and Electrical elements. Using the 

multi-user environment allows the various departments to 

combine their efforts in a single Project environment, 

sharing information and being dynamically kept up to date 

with any changes which affect the design. 

Figure 7: Combined finished article. 

5. FUTURE DEVELOPMENTS 

5.1 CONCEPT TO DESIGN APPROACH 

This technology is very much in its infancy; the uses 

discussed above are very much the obvious and have

een driven by user requests for a more tightly integrated 

Electrical/Pneumatic design space and to allow companies to 

flood Projects with engineers. However, we are already 

starting to see companies exploring the possibility of using 

the functionality to adopt a more ‘Top Down’ system design 

philosophy. Concept Engineers, with very much an overview 

of the design requirements are placing building blocks within 

a Project. These building blocks are embellished further by 

Systems Engineers, adding function to the blocks and then 

finally Design Engineers are producing detailed schematics 

of the systems. 

The linkage between these three levels is yet again dynamic, 

with each department being able to update and modify their 

designs, with any change being automatically reflected in the 

other levels. This methodology could be described a 

Hierarchical design approach. An example of this would be 

the building blocks which go into the design of a new vehicle. 

Conceptually there will be two doors, four seats, two 

headlights etc, these blocks are placed into the design. On 

an alternate view the System Engineer shows the simple, 

single line connections between these blocks. How they are 

related and interfaced, then finally the Design Engineer has 

the job of filling in the detail. How the system is wired up, all 

79 

the contacts and switches. Here again all departments 

have the opportunity to look and check each other 

departments ideas, concepts, technology. Mistakes can 

be spotted and rectified and by bringing different 

departments into a shared design space, new ideas could 

be formulated. 

6. SUMMARY 

In the abstract for this paper it was hoped that by writing 

this paper new ideas could be formulated for the use of a 

collaborative ECAD tool an in deed a number of concepts 

have come to light, namely the quality aspect. But it is felt 

that there are many more uses which will fall out of this 

technology and they will predominantly come from the 

ECAD engineers desire to achieve better tools for their 

current practices.

ECAD/ECAE2004 

ECAD/ECAE2004 

1st International Conference on Electrical / Electromechanical Computer Aided Design & Engineering 

15-16 15 16 November 2004, Durham, UK 

Electrical & Process Automation Framework 

for 

Engineering and Maintenance 

with 

3rd CAE- CAE Systemgeneration 

Technische Computer Systeme 

Süssen GmbH 

SIGRAPH CAE 

Dipl.-Ing. Dipl. Ing. 

Debasish Mukherjee 

Director International Business 

80

Contents 

■ What is it all about? 

■ Modules of Electrical and Process Automation Framework 

■ User Benefits 

■ Electrical and Process Automation Workflow 

/ SIGRAPH CAE 

Two names, one product 

■ Electrical and Process Automation Data in One Engineering Database 

■ Data exchange within Electrical and Process Automation Framework 

■ Technology and Product Life Cycle 

81

What is it all about? about 

Electrical & Process Automation 

✓ Electrical and I&C 

Engineering- Engineering and Maintenance-Framework 

Maintenance Framework 

✓ Modular Solution for the whole Life-Cycle Life Cycle of Machines and Plants 

3. CAE-Systemgeneration 

CAE Systemgeneration 

✓ New development middle/End middle/End 

of 90 

✓ New Target as 2nd System Generation 

✓ CAEngineering 

CAEngineering 

as base of one Engineering Database 

✓ Introduction of latest Technology - Integrated Object Orientation 

SIGRAPH CAE 

82 

/ SIGRAPH CAE 

Two names, one product

Framework for Electrical & Process Automation 

Basic-Engg 

Main Modules of Engineering & Maintenance 

Multi tasking Concurrent 

Engineering 

Electrical & Process Automation framework for Engineering and Maintenance 

Leads to 

Electrical & Process Automation workflow for Engineering and Maintenance 

Main Functions 

Detail- 

Engineering 

PLC/ Controls Documentation 

and Revesion 

Management 

Customer 

Oriented 

User- 

Interface 

ERP 

Report / 

Online-List 

/ SIGRAPH CAE 


Engineering-Database 

Engineering Database 

Object Oriented 

EDM / PDM PLC 

Modules for System Integration 

Viewing/ 

Navigation 

• Daten-Browser 

• PDF 

... 

... ... 

83

User Benefits 

Support of 

Electrical & 

Process 

Automation 

Workflow 

Electrical and 

Process 

Automation Data 

in one 

Engineering 

Database 

Electrical and Process 

Automation Framework 

for 

Engineering & Maintenance 

SIGRAPH CAE 

New 

Data 

Technology Exchange 

Begin of 

the product 

Life Cycle 

84 

System- 

integration 

/ SIGRAPH CAE 


Electrical and Process Automation-Workflow 

Automation Workflow 

✓ Global support for Organisation as well as Order 

specific Workflows 

✓ Is applicable to the complete Life cycle of 

Equipment and Machine 

✓ In moulding phase itself all data are inputted in 

the Engineering Database 

✓ Data are available at any point of time to any 

Engineering stage for further processing 

✓ Jobs can be executed in parallel 

✓ Beginning of the Engineering activity is not 

limited to circuit diagram only 

85 

/ SIGRAPH CAE 


SIGRAPH CAE 


Electrical and Process Automation Data in Engineering Database 

Flow Chart 

Engineering of 

Measurements, 

Measurements, 

Instrumentation, 

Instrumentation, 

E-Consumers Consumers etc 

General 

Arrangement 

Drawing 

Loop plans 

Engineering Schranklayou 

Database of 

t, -disposition disposition 

Electrical & Process 


Single-Lines 

Single Lines 

86 

Engineering 

of PLC/ 

Controls 

Cable Planning 

Engineering of 

Power Supply 

Generation of 

Circuits 

Diagrams

SIGRAPH CAE 


Dataexchange within Electrical & Process Automation Framework 

Consultant 

Consultant 

Engineering 

Engineering 

(Operator) 

(Operator) 

Electrical and Process 

Automation-Framework 


for 

Engineering & Maintenance 

Engineering-Data 

Engineering Data bank 

Object Oriented 

87 

Designer 

Designer 

Maintenance 

Maintenance 

(Operator) 

(Operator) 

Manufacture, 

Manufacture 

Manufacture, 

Commissioning, 

Commissioning 

Commissioning, , 

Service 

Service

Technology und Product Life-cycle Life cycle 

/ SIGRAPH CAE 


3rd CAE-System 

CAE System Generation is already established in the Engineering World ... 

✓ Integrated Object Orientation (Database, Implementation, Implementation, 

User-Interface) 

User Interface) 

✓ Available software architechture for Integrated and phasewise Engineering 

✓ New Demands of the Market since middle of 90s 

✓ Beginning of Product Life-Cycle Life Cycle 

New product for the entire 

Market, ensures long term 

Investment security ! 

88 

Bedarf 

Forum Factory Automation 2004; 22.04.04 

/ SIGRAPH CAE 

Zwei Namen - Ein Produkt 

Lebenszyklen 2. ECAD-Generation / 3. CAE-Generation 

Lebenszyklus 

2. ECAD- 

Systemgeneration 

1985 1995 1999 / 2000 2004 2009 / 2010 

Start Entwicklung 


Markteinführung 


Umstiegszeit 

ECAD CAE 

Lebenszyklus 

3. CAE- 

Systemgeneration 

Zeit in Jahre 

10

Technische Computer Systeme Süssen GmbH 

Tobelstraße 8 

73079 Süssen 

Germany 

www.tcs-s.de 

www.tcs s.de 

tcs@tcs-s.de 

tcs@tcs s.de 

Tel.: +49 (0) 7162 / 4095 - 0 

Fax: +49 (0) 7162 / 41032 

Debasish Mukherjee 

Director International Business 

d.mukherjee@tcs-s.de 

d.mukherjee@tcs s.de 

- Thinking in Solutions 

89 

/ SIGRAPH CAE 


Solutions! Thank you ve 

ry much 

for your attention

Automatic Wiring in Switch Cabinets 

W.Vigerske, B.Stube and M.Pleßow 

R&D Department Graph-based Engineering Systems 

Society for the Promotion of Applied Computer Sciences (GFaI), Berlin, Germany 

Abstract 

The computer-aided design of switch cabinets opens the possibility to automate certain layout steps and to 

get an optimal layout by efficient, dedicated tools. Among the typical layout steps automatic wiring of 

switch cabinets has the most relevance and is a great challenge. Beside the fact that an automatic wiring 

delivers an entire (if possible), valid and optimal result, an additional important aspect is the possibility to 

score customizing cables. Subject of this article are mathematical problems in the field of automatic wiring 

in switch cabinets. 

Keywords: 

ECAD, Layout, Wiring 


Switching stations can be found in modern factories and 

buildings everywhere. They are utilized as supervisory 

station to control and observe different technical 

processes. 

ECAD-systems are often used to design switchboards. 

As design result circuit diagrams are mainly developed. 

Based on a circuit diagram the needed connection list 

and the device list can be derived. It is typical for ECADsystems 

that they cover all relevant facets of the design 

process which concern the logic structure of the 

switchboard. Such systems have a long history and offer 

a wide range of comfortable features to design circuit 

diagrams. But for the physical design of a switchboard 

they offer no assistance. This fact is the reason of the 

strong demand for systems which are able to realize an 

integrated approach for handling of general process 

chains [1]. 

The most important layout step during switchboard 

design is wiring of the connection structure. Automatic 

wiring tools are mainly based on classical graph 

theoretical approaches. But the relevant switchboard 

applications require a significant extension of the 

existing strategies. Since many of these layout problems 

are NP-complete new efficient heuristics have to be 

developed. 

2 AUTOMATIC WIRING 

A switchboard consists mainly of different switch 

cabinets, which contain mounting plates for the electrical 

devices. For connecting electrical devices, conductors 

are used and laid out through a channel framework (see 

Figure 1). 

The aim of automatic wiring is to compute a short and 

cost optimal total wiring respectively. During wiring 

structural aspects have to be considered, which are 

given by the resolution process of T-connections 

(electrical connections of more than two terminals) in 

order to realize the connecting structure in the switch 

cabinet. Furthermore, wiring is influenced by the facts 

that cable channels have a limited capacity and 

sometimes constraints to avoid electromagnetic 

compatibility (EMC) problems have to take into account. 

So, the degrees of filling of the cable channels have to 

be modelled and considered during wiring computation. 

90 

Figure 1: Loaded mounting plate. 

In the next chapters we will describe relevant aspects for 

the generation of a single connection in the channel 

framework. This includes the characterization of the 

connecting problem as a shortest path problem in an 

edge-weighted graph, the graph generation based on the 

channel framework and the degree of filling issues. 

2.1 Optimal wiring of a single connection 

From the algorithmic point of view the computation of an 

optimal total wiring in the channel framework is a NPcomplete 

problem [2]. Even the optimal realization of a 

single net of the circuit diagram, the resolution of a Tconnection, 

is probably NP-complete. So, for a practical 

handling of the wiring problem there is the orientation to 

resolve the T-connections sequentially. Also a single Tconnection 

will be resolved by sequentially generation of 

optimal single connections. For the optimal layout of the 

single connections the well known Dijkstra algorithm [2] 

can be applied. This algorithm computes cost minimum 

paths in a graph with positive edge-weights between 

start and destination nodes. With regard to the resolution 

problem of T-connections the algorithm is modified and 

extended. 

2.2 Preparation of the wiring problem 

For the application of the Dijkstra algorithm the needed 

edge-weighted graph has to be extracted from the 

channel framework. Basis for this graph generation is 

the channel model. In this model the information about

the location of the cable channels, their dimensions and 

structuring are mapped. Structuring includes the 

information which of the cable channels have transitions 

and where are they located. Based on this channel 

model a static graph is generated. 

For wiring of a single T-connection all relevant terminals 

will be integrated in the static graph. By this means the 

routing graph is constructed (see Figure 2). Based on 

this routing graph the T-connection layout will be 

computed as a sequence of pair-wise connections 

between the terminals. 

Figure 2: Routing graph with entry points derived from 

different terminals 

Important and not simple is the modelling of the filling 

degree of the cable channels. For this modelling each 

cable channel has assigned a list of free areas. In each 

wiring step the channel capacity will be adapted in this 

list. 

2.3 Resolution of T-connections 

During wiring of a T-connection in a switch cabinet the Tconnection 

will be resolved in pair-wise connections 

between the terminals of the electronic devices which 

will be physically realised by electrical conductors while 

the fabrication of the switch cabinet. Figure 3 shows the 

T-connection -X1:1, -F13.1:2, -K27.1:6 which is part of a 

circuit diagram. 

Figure 3: T-connection example. 

The realization of this T-connection in a switch cabinet is 


X1 

1 

2 

K27.1 

3 

F13.1 

Figure 4: T-connection realization. 

In this resolution process it is not important which pairwise 

connections will be really realised. But, it has to be 

guaranteed that in the set of the given terminals there 

exists an electrical connection between any two 

terminals and the capacity of the terminals allows the 

envisaged connection sequence. In this example 

terminal X1:1 has capacity 1. Therefore, terminal X1:1 

can only be used as first or last terminal of a connection 

sequence. 

A T-connection can be considered as a hypernet and the 

key problem of routing is the decomposition of this 

hypernet. The terminals involved in a hypernet must be 

connected by a tree, which branches out only at the 

terminals. So the decomposition process of a hypernet 

(T-connection) involved the following problems: 

• It is not sufficient to route pair-wise connections 

optimal. Optimal tree structures have to be 

computed. 

• Terminals have maximum valences for technical 

reasons. It has to be clarified whether a 

decomposition of the hypernet is possible. 

Furthermore, since the tree structure will be serially 

realised by pair-wise connections there is the 

question which connection sequence is admissible 

that the procedure will not stopped by capacity 

reasons of the terminals during computation. 

2.4 Optimal tree in an edge-weighted graph 

We consider all terminals as nodes in a complete graph 

V . The capacity of the terminals will be characterized by 

+ 

valences. For a node v ∈ V let potval ( v) 

∈ N the 

given prospective valence. This valence is the maximum 

number of edges (connections) which can incident in this 

node. Let val( v) 

∈ N0 

be the actual number of incident 

edges in node v and freeval( v) 

= potval( 

v) 

− val( 

v) 

is 

the free connection capacity. 

In the case of potval( 

V ) = ∑ potval( 

vi 

) ≥ 2( 

n −1) 

than 

i= 

1 

a set of edges E can be generated that T = ( V , E) 

is a 

spanning tree (a resolved T-connection) in the complete 

graph V . This tree meets the valence constraints. For 

the spanning tree generation only the information about 

a global connection capacity is needed. The exact 

distribution of this global capacity among the nodes is 

not important. The constructive proof of this fact is also a 

recommendation for the generation of a valid connection 

sequence. The constellation at the beginning is the pure 

set of nodes of graph V . By successive addition of 

edges we generate subtrees T i . In the case that more 

than one subtree exist we have take into account that 

91 

n 

KK1

criteria freeval ( Ti 

) ≥ 1 is fulfilled. Is there only one 

subtree, we have generated an envisaged spanning tree. 

2.5 The problem of multiple T-connections 

Two paired T-connections T1 and T2 are shown in Figure 

5. T-connection T1 represents a power subcircuit and 

connects terminals E, G, F and D. T-connection T2 

represents a control unit with terminals A, B, C and with 

one or more connections to T1. 

10 

T 1 

G F D 

E 

2.5 

T 2 

A B C 

Figure 5: Multiple T-connections. 

The connection from T2 to T1 is not more specified. In 

the resolution process of T1 all terminal capacities of E, 

i 

G, F, D are used. We define d X as capacity which is 

consumed by terminal X during the resolution of Tconnection 

Ti. Than is valid: 

1 

1 

1 ≤ d E ≤ potval( 

E) 

, 1 ≤ dG ≤ potval( 

G) 

, … 

1 1 1 1 

and d G + d E + d F + dD 

= 6 . 

We can consider the T2 resolution as computation of 

connection structure between terminals A, B, C and a 

virtual terminal T1 (Figure 6). 

T1 A B C 

Figure 6: T-connections virtual terminal. 

In this case is also valid: 

2 

1 ≤ d A ≤ 

2 

potval( 

A) 

, 1≤ 

d B ≤ potval( 

B) 

, … 

2 

1 ≤ dT ≤ potval( 

T1) 

, 

d 

1 

+ d 

+ d 

+ d 

= 

2 2 2 2 

A B C T1 

and 

. 

2 2 2 2 2 2 2 

= d A + d B + dC 

+ dG 

+ d E + d F + d D = 6 

1 2 

Inequalities like d A + d A ≤ potval( 

A) 

and so on have 

to be fulfilled as well. By this means we have got a 

system of diophantic equations and inequalities (DiGL). 

For each T-connection resolution a set of terminal 

capacities can be reserved. If all capacity sets meet the 

system DiGL than all T-connections can be resolved. For 

each T-connection the set of terminal capacities can be 

92 

considered as a restriction for possible connection 

structures. 

2.6 Construction of a minimum connecting 

structure 

Now, in order to compute cost minimum trees we assign 

cost to each edge in graph V . The well known Kruskalalgorithm 

[4] can not be applied because it only works on 

graphs without restrictions concerning node valences. 

Therefore, we have developed a novel heuristic to solve 

this problem. This approach is similar the strategy we 

use for serial construction of optimal trees. 

The algorithm works on an edge-weighted graph 

G = ( V , E) 

and a set of nodes K ⊆ V . We will construct 

a minimum spanning tree in K with edges which belong 

to graph G . 

1. All nodes of subset K are defined as (trivial) 

subtrees. 

2. Are there subtrees T with freeval ( T ) = 1 , we assign 

such trees to the start set. If not, than the tree 

selection for the start set is arbitrary. Let subtree s T 

be selected. All nodes k of T s with freeval ( k) 

> 0 

define start set S . 

3. Are there more than two subtrees all subtrees T z 

with freeval ( Tz 

+ Ts 

) > 2 will be selected. Otherwise 

all remaining subtrees can be used for the 

destination set. All nodes k of set T z with 

freeval ( k) 

> 0 are used for destination set Z . 

4. Start of the modified Dijkstra-algorithm. A cost 

minimum path will be computed which connects a 

node of start set S with a node k zz of destination 

set Z ( k zz belongs to i T ). By this s T and T i will be 

merged to a new subtree. 

5. Is there more than one subtree than go to 2 

6. STOP. 

Now, we will roughly explain the applied modified 

Dijkstra-algorithm using an example (see Figure 7). 

1 

2 

9 

19 

19 

7 

18 

39 

34 

3 

4 

5 

6 

7 

Start nodes = 1,2; Destnation nodes = 17,18,19 

Cost minimum path = 2,5,11,14,17 with costs = 36 

33 

12 

8 

11 

16 

12 

35 

20 

32 

48 

8 

11 

9 

10 

13 

44 

22 

13 

2 

37 

21 

6 

12 

13 

14 

15 

16 

27 

20 

4 

3 

31 

1 

9 

18 

28 

7 

6 

47 

14 19 

Figure 7: Modified Dijkstra algorithm. 

In the example nodes 1 and 2 are start nodes (set S) 

and nodes 17, 18 and 19 are destination nodes (set Z). 

We compute the shortest path between each start and 

destination node. For each combination 

( vi, v j ), vi 

∈ S, 

v j ∈ Z 

a shortest path exists. Based on the 

11 

17

set of all shortest paths the algorithm finds the best 

combination (start node, path, destination node) with 

minimum costs without computation of all combinations. 

For n nodes in subset K the computation of (n-1) 

shortest paths is necessary. We can estimate the 

algorithmic complexity with O(|V| 2 *|K|). The algorithm 

used a Greedy-strategy. This means that only the next, 

absolute minimum path is taken. So, suboptimal paths 

which would further reduce the shortest path length are 

not considered. Therefore, the algorithm can not find the 

optimal solution in the following example (Figure 8). 

... ... 

... 

Figure 8: Example for a not optimal solution. 

As you can see all terminals have a capacity of 2. The 

algorithm will begin with the left most terminal in the 

upper row as start node and will ever connect the next 

terminal to the right. Unfortunately, in the final step a 

connection between the terminal most to the right and 

the terminal in the second row will be generated. In the 

optimal solution the terminal in the second row would be 

connected using the vertical channel (dotted line). But in 

realistic cases the algorithm delivers good results. 

2.7 Cable routing 

The basic algorithm for automatic wiring was extended to 

handle the problem of cable routing. This routing 

problem can be described as followed: A given shielded 

cable consists of several insulated conductors. At both 

cable ends it has to be spliced that the different wires 

can be connected with the corresponding terminals. So, 

the task of optimal cable routing consists in to compute 

optimal splice points with the restriction that the wires 

are connectable with the terminals and the unshielded 

wire lengths are minimal. We have developed a novel 

algorithm which computes both optimal splice points and 

an optimal wiring of the unshielded connections. The 

algorithm determines the optimal spanning tree 

connecting the terminal points. Due to the rectangular 

layout of the channel framework the optimal splice point 

can be calculated by walking through the spanning tree 

and to register the balance of terminal points "behind" 

and "in front" of the current position in the spanning tree 

[5]. 

3 SUMMARY 

It has been shown that the developed wiring algorithms 

can meet the practical requirements. They are integrated 

in an ECAD-system and used for wiring of mounting 

plates. The next work is concentrated on wiring of a 

system of mounting plates. 

... 

Figure 9: Wired mounting plate 

4 ACKNOWLEDGEMENT 

The authors are grateful to the Federal Ministry of 

Economics and Labour of Germany (BMWA) for the 

financial support of several projects on this subject such 

as KAVal under the project code 1024/99 and Autent 

under the project code KF 0012518KWD2. 

5 REFERENCES 

[1] Schneider, D., Roller, D., 1999, Elektro-CAD am 

Wendepunkt, CADWORLD 5. 

[2] Lengauer, T., 1990, Combinatorial algorithms for 

integrated circuit layout, John Wiley&Sons, 

[3] 

Chichester, UK. 

Vigerske, W., Goetze, B., Pleßow, M., Wrobel, G., 

1999, Automatische Verdrahtung in Schaltanlagen, 

ZwF Zeitschrift für wirtschaftlichen Fabrikbetrieb, 

Heft 1 und 2, Carl-Hanser-Verlag, München. 

[4] Kruskal, J.B., 1956, On the shortest spanning 

subtree of a graph and the travelling salesman 

problem, Proceedings of the Amarican 

[5] 

Mathematical Society, 7(1), pp. 48-50. 

Vigerske, W. :Diskussionspapier zum Kabelrouting. 

Arbeitsdokument, GFaI-Autent-22-03, GFaI, Berlin, 

2003. 

93

PART 5 

PCB DESIGN & KNOWLEDGE 

MANAGEMENT 

94

Abstract 

Hybrid Genetic Algorithm for Xilinx-style FPGA Placement 

M. Yang and A. E. A. Almaini 

School of Engineering 

Napier University, Edinburgh, EH10 5DT 

Genetic algorithm (GA) techniques are well -suited for solving a wide range of combinatorial optimization 

problems. However, infeasible individuals may be generated during the search in the constraint search 

space. This paper introduces a useful representation for two-dimensional Xilinx-style FPGA placement. To 

achieve fast convergence and better solution, a hybrid GA is used. Empirical results show fitness 

improvement of 52% on average on 9 MCNC benchmarks com pared to the results using standard GA based 

on 200 generations . After placement, the competitive routing channel results are compared to the res ults 

obtained by the Versatile Place and Route (VPR) package. 

Keywords: 

Hybrid Genetic Algorithm, Placement, FPGA 


Since their first introduction in early 80s, Field 

Programmable Gate Arrays (FPGAs) have gained 

increasing popularity in implementing digital circuits. 

Especially as process geometry has shrunk to deep- 

submicron, the logic capacity of FPGAs has significantly 

increased. In FPGAs, all routing resources are 

prefabricated; the width of all the routing channels is set 

by the FPGA manufacturer. The goal of placement and 

routing is then to find efficient utilization of limited and 

prefabricated routing resources for different types of 

circuits. 

Placement problem has been found to be NP-complete, 

implying there is unknown deterministic algorithm to solve 

it in polynomial time. Several algorithms such as min-cut 

algorithm [1], force-directed algorithm [2], and simulated 

annealing [3] and so on have been proposed to solve this 

problem. 

Genetic algorithm (GA) is stochastic search algorithm 

based on biological evolution models, whose main 

advantage lies in its robustness of search and problem 

independence. The basic concepts of GA were developed 

by Holland in 1975 [ 4]. Since then it has been widely 

applied for solving problems in physical layout, including 

standard cell placement [ 5], macro cell placement and 

routing [6], traditional channel routing [7] and over-the-cell 

(OTC) routing [8]. 

95 

Although GA has characteristics of robustness and wide 

range search space, it normally takes a large number of 

generations to converge to optimum. This makes fast 

prototyping of FPGA difficult and impractical for large 

circuits. As a result, a hybrid algorithm is introduced to 

overcome long computation time. 

In the paper, section 2 gives a brief description of FPGA 

architecture, in particular symmetrical architecture used in 

the experimental results . Section 3 presents placement 

problem definition in Xilinx-style FPGA. In section 4 

description of hybrid genetic algorithm is given. 

Experimental results showing the effectiveness of the 

proposed approach are presented in section 5. 

Conclusions are then given in section 6. 

2 FPGA ARCHITECTURE 

In 1985 Xilinx Inc introduced the first Look-Up Table 

(LUT) based FPGA. FPGAs are more flexible and 

complex than other programmable devices such as 

Programmable Logic Array (PLA) and Programmable 

Array Logic (PAL). With the rapid improvements in the 

performance and logic densities of the FPGAs , the 

number of applications continues to increase. 

An FPGA consists of configurable logic blocks (CLBs), 

which typically contain either combinational or sequential 

logic circuits, Input/Output blocks (IOBs) and routing 

resources such as wire segm ents and programmable

switches [9]. Programmable switches configure the wire 

segments between logic blocks and between logic and 

I/O blocks. Over the last 10 years, several companies 

have introduced a number of different types of FPGAs 

based on much more complicated architecture than 

previous FPGAs to improve their performance. The 

architecture of an FPGA can be subdivided into two main 

parts, logic block architecture and routing architecture. 

The logic block architectures [10] are: 

1. Look-up tables (LUTs); 

2. multiplexers; 

3. PLD blocks ; 

4. transistor pair 

The routing architecture is an important characteristic of 

an FPGA as routing resources, which occupy 50-90% of 

the chip area [11]. All types of routing architectures [10] 

can be classified into one of four categories . These are: 

1. symmetrical; 

2. row based; 

3. sea-of-gates ; 

4. hierarchical 

In this paper, we will exclusively investigate the 

symmetrical routing architecture. This style of routing 

architecture is shown in Figure 1. 

Figure 1: General architecture of Xilinx FPGA. 

Logic blocks are surrounded by routing channels of 

prefabricated wiring segments on all four sides. A logic 

block input or output, which is normally called a pin, can 

connect to some or all of the wiring segments in the 

channel adjacent to it via a connection block of 

programmable switches. At every intersection of a 

horizontal channel and vertical channel, there is a switch 

block. This is simply a set of programmable switches that 

allow some of the wire segments incident to the switch 

block to be connected to others. By turning on the 

96 

appropriate switches, short wire segment can be 

connected together to form longer connections. 

3 PLACEMENT PROBLEM DEFINITION 

The goal of placement algorithm is to determine which 

logic block within an FPGA should implement each of the 

logic blocks required by the circuit. The optimized 

placement is to place connected logic blocks close 

together so that the wire length of placement can be 

minimized. 

In this paper, we only focus on one optimization goal of 

minimizing routing channel density (routability-driven 

placement). As a result, the quality of the placement 

depends on the routing density after the router connects 

prefabricated programmable switches while implementing 

the circuit. 

4 HYBRID GENETIC ALGORITHM 

4.1 Genetic algorithm 

The genetic algorithm (GA) is a search technique which 

mimics the natural process of evolution as a means of 

progressing to the optimum. It starts with an initial set of 

random configurations termed a population. Each 

individual in the population is a string of genes. The string 

made up of genes is termed chromosome. The 

chromosome represents a solution to the problem. During 

each iteration, which is termed generation, the individuals 

in the current population are evaluated using 

measurement of fitness function. Based on the fitness 

value, two individuals at a time are selected from the 

population. The individuals with higher fitness value will 

have more chance to be selected by using crossover 

operator followed by mutation operator to generate new 

individual solutions called offsprings. 

Hybrid GA (HGA) is a standard GA (SGA) which performs 

local optimization in every generation to overcome long 

computation time and improve fitness of SGA. The HGA 

is shown in Algorithm 1. The algorithm begins with an 

initial set of random population. After evaluating fitness of 

current population, the population is reproduced 

according to fitness. The fitter the individual, the more 

chance it has to be selected. Two individuals are 

randomly selected as parents to generate offsprings by 

using crossover operator based on high probability of 

crossover. Mutation operator with low probability rate is 

carried out. After that, local improvement with low

probability rate is applied to randomly selected individuals 

so that visible improvement can be achieved, resulting in 

shorter search time. The elitism is employed to retain the 

good solution. After a fixed number of generations, the 

fittest individual, namely the one with highest fitness 

value, is returned as the desired solution. 

MAX_GENS: maximum number of generations 

POP_SIZE: population size 

NUM_GENE: number of genes 

NUM_BLOCK: number of blocks for each benchmark 

NUM_MOVE: number of moves per individual in local 

improvement 

Pcrossover: probability of crossover rate 

Pmutation: probability of mutation rate 

Plocal: probability of local improvement rate 

begin 

Generate an initial population 

for generation = 1 to MAX_GENS do 

end for 

end algorithm 

evaluate population fitness values; 

reproduce population probabilistically 

based on the individual's fitness value; 

for i = 1 to POP_SIZE/2 do 

pair two parents randomly; 

crossover based on Pcrossover; 

produce two new offspring; 

end for 

for j=1 to NUM_GENE do 

end for 

mutate offsprings based on Pmutation; 

for i = 1 to POP_SIZE do 

local improvement based on Plocal and 

NUM_MOVE ; 

end for 

elitism; 

Algorithm 1 : Pseudo-code of HGA. 

4.2 Genetic encoding 

In the above GA procedure, a chromosome pertaining to 

a possible placement solution is represented as a string 

with length which equals to the numbers of logic blocks in 

the FPGA. For example, if the size of FPGA is 4 by 4, the 

numbers of the blocks are 16 and the length of 

97 

chromosome is 16. The values of string can be either 

positive integer or -1. The positive integer represents ID 

of the block and the value -1 represents empty block. The 

position of block in an FPGA is numbered according to its 

X-Y position. The block ID is mapped to chromosome 

according to its number, e.g. Block 10 with number 1 in 

the FPGA is mapped to position 1 of the chromosome as 


4.3 Selection operator 

Figure 2: Genetic encoding. 

The GA procedure carries out the genetic selection 

operator in which individual strings are chosen according 

to their fitness values. A proportional selection scheme as 

suggested by Goldberg is employed to select fitter 

parents which are required for reproduction. There are a 

number of ways to implement the selection operator. The 

easiest way is to create a biased roulette wheel where 

each current chromosome in the population has a roulette 

wheel slot sized in proportion to its fitness. An individual is 

selected by spinning the roulette wheel and noting the 

position of the marker. However, the absolute difference 

between an individual’s actual sampling probability and its 

expected value is nonzero, resulting in inefficient 

reproduction. Hence stochastic universal selection with 

zero bias [13] is em ployed in the reproduction process 

4.4 Fitness measure 

A fitness function is used to evaluate the quality of 

placement. Its functional form is the sum of all nets in the 

circuit, as shown in Equation 1 

f = 

N 

100 

∑ q(i)[bb (i) + bb (i)] 

x y 

i = 1 

where N stands for number of nets. For each net i, bbx (i) 

(1)

and (i) 

bb y 

denote the horizontal and vertical spans of its 

bounding box respectively. q(i) factor, which is adapted 

from [12], compensates for the fact that the bounding box 

wire length model underestimates the wiring necessary to 

connect nets with more than three terminals. Its value 

depends on the number of terminals of the net i. 

It should be noted that high fitness value indicates a 

placement with shorter wire length, hence a better 

solution. 

4.5 Crossover operator 

Crossover is the main genetic operator. It operates on two 

individuals and generates two offsprings. It is an 

inheritance mechanism where the offspring inherits some 

of the characteristics of the parents. The operation 

consists of choosing a random cut point and generating 

the offspring by combing the segment of the other parent 

to the right of the cut. Unfortunately, the elements to the 

left of crossover cut point in one parent appear on the 

right of the second parent. This duplication does not 

represent a feasible placement solution. Modification of 

crossover to avoid duplication has to be carried out. 

The modification is implemented as follows. Choose a 

random cut point and copy the entire segments following 

the cut point in parent 1 to the offspring. Next, the left 

segment of parent 1 is scanned from the left most, gene 

by gene, to the point of the cut. If a gene does not appear 

in the offspring then it is copied to the offspring. However, 

if it already exits in the offspring, then its position in parent 

2 is determined and gene from parent 1 in the determined 

position is copied. If the determined gene still exits in the 

offspring, determine the position in parent 2 as before 

until the gene does not exit in the offspring. One particular 

case is the gene -1 which means the block is empty. If 

there is no empty block in parent 2, the empty gene is 

copied to the offspring in the same way as a gene does 

not appear in the offspring. However, if empty blocks do 

exit, it will randomly select any one of empty positions in 

parent 2 and the gene from parent 1 in the selected 

position is then copied to the offspring. The selected 

empty position is marked to avoid being us ed again. An 

example is shown in Figure 3. One possible placement, 

which is shown in Figure 2, is encoded as a string of 16 

bits, namely, { -1,10,2,-1,1,3,8,5,-1,4,-1,6,9,7,-1,11}. The 

other parent is {1,-1,-1,3,7,11,10,-1,5,6,8,9,-1,2,4,-1}. The 

cut point is 7. 

98 

4.6 Mutation operator 

Figure 3: Modified crossover. 

Mutation produces incremental random changes in the 

offspring generated by the crossover to overcome early 

converge to a local optimum . In the placement, the 

mutation is pair-wise interchange, namely, two genes of 

the chromosome are randomly selected according to 

probability of mutation rate and their positions swapped. 

4.7 Local improvement 

After reproduction, crossover and mutation are 

performed, a local improvement is applied to the selected 

offspring in the current population based on the 

probability of local improvement rate. The local 

improvement is performed in every generation and keeps 

switching blocks in the FPGA a number of times for 

randomly selected individual in order to improve the 

fitness of this particular individual. Some good schema of 

this individual is more likely to be selected and passed to 

next generation. The main goal of this process is to get 

some visible improvement in the offspring rather than 

obtaining a local optimum value. So the improvement rate 

is kept as low as possible and movement per individual is 

also kept to small value. As a result, the time for 

convergence is reduced significantly. 

5 EXPERIMENTAL RESULTS 

In this section, experimental results obtained with an 

implementation of SGA and HGA to Xilinx-style FPGA are 

reported. The implementation is written in the C 

programming language. All experiments were performed 

on Intel Celeron 897 with 128M memory under Linux. 

Performance is measured using 9 MCNC benchmark 

circuits from VPR placement and routing tool [12]. (Note: 

the authors can not find benchmark alu4 so e64 which 

was shown in the appendix of [12] is used instead.) 

Table 1 lists the main characteris tics of these benchmark 

circuits.

Name 

No. of 

blocks 

No. of 

nets 

No. of 

CLBs 

No. of 

I/Os 

9symml 107 106 97 9/1 

alu2 213 207 197 10/6 

apex7 188 151 102 49/37 

e64 404 339 274 65/65 

example2 289 223 138 85/66 

k2 609 564 519 45/45 

term1 132 122 88 34/10 

too-lrg 228 225 187 38/3 

vda 374 308 291 17/39 

Table 1: Benchmark characteristic 

The parameter values are selected following some 

experiments and based on previous experience. 

Pcrossover=0.6, Pmutation=0.005, POP_SIZE=50. Since 

SGA does not involve local improvement mutation 

operator, the parameter values are Plocal=0 and 

MAX_GENS=1000. While in HGA, the parameter values 

are Plocal=0.05, MAX_GENS=200 and NUM_MOVE=10 

X NUM_BLOCK. For example, NUM_MOVE =1070 for 

9symml in Table 1. 

As can be seen in Table 2, the fitness is improved on 

average by 79% in 1000 generations compared to initial 

random cost of placement. Further 52% improvement just 

in 200 generations is gained when HGA is employed 

compared to SGA. Figure 4 illustrates initial fitness before 

optimization, fitness obtained by SGA and HGA for 9 

MCNC benchmarks, respectively. 

The placement from VPR and our SGA and HGA 

algorithms are routed by VPR router. The resulting 

channel densities are compared, as shown in Table 3. 

The main steps are illustrated by the flow chart in 

Figure 5. All parameter values for routing are set equal for 

the purpose of fair comparison. 

Name 

initial 

average 

fitness 

f 

fitness 

of SGA 

f 

improvement 

comp. to 

init.fitness 

% 

fitness 

of HGA 

f 

improve 

-ment 

comp.to 

SGA 

% 

9symml 1349 2147 59 2799 30 

alu2 1732 2921 69 4159 42 

apex7 1495 3299 121 4364 32 

e64 2246 3638 62 6391 76 

example2 1528 3626 137 6746 86 

k2 3062 4749 55 8261 74 

term1 1323 2375 80 2944 24 

too-lrg 1957 3239 66 4557 41 

vda 2033 3258 60 5293 62 

average 79 52 

Table 2: Comparison of fitness 

99 

fitness 

10000 

9000 

8000 

7000 

6000 

5000 

4000 

3000 

2000 

1000 

0 

9symml 

alu2 

Comparison of fitness 

apex7 

e64 

example2 

benchmark 

k2 

term1 

initial average 

fitness 

fitness obtained 

by SGA 

fitnest obtained 

by HGA 

Figure 4: Comparison of fitness based on Table 2. 

Placer FPGA 

SGA HGA 

VPR 

Router Size 

VPR VPR 

9symml 10x10 5 6 5 

too-lrg 

alu2 15x15 6 10 7 

apex7 11x11 5 8 5 

e64 17x17 8 17 8 

example2 19x19 5 8 5 

k2 23x23 9 20 11 

term1 10x10 5 7 5 

too-lrg 14x14 7 12 8 

vda 18x18 8 14 9 

Total 58 102 63 

Table 3: Comparison of channel density of 

GA and VPR 

Figure 5 : CAD flow chart. 

vda

6 CONCLUSIONS 

In the paper, SGA and HGA for Xilinx-style FPGA 

placement are presented. The experiment verified that the 

proposed SGA is effective for improving fitness for small 

size circuits. The fitness improved by 79% on average 

compared to fitness without optimization for 9 MCNC 

benchmark circuits , though a large number of generations 

is required to convergence to optimum. However, the 

proposed HGA can overcome this problem. It can obtain 

further improvement of 52% on fitness on average 

compared to the results obtained by SGA and significantly 

reduce the numbers of generations for convergence. 

7 REFERENCES 

[1] Lauther, U., 1979, A min-cut placement algorithm of 

general cell assemblies based on a graph 

representation, Proceedings of 16th Design 

Automation Conference, 1-10. 

[2] Johannes, F. M., Just, K. M. and Antreich, K. J., 

1983, On the force placement of logic arrays, 

Proceedings of the 6th European Conference on 

Circuit Theory and Design, 203-206. 

[3] Sechen, C. and Sangiovanni-vincentelli, A., 1986, 

TimberWolf3.2: A new standard cell placement and 

global routing package, Proceedings of 23rd Design 

Automation Conference, 432-439. 

[4] Goldberg, D. E., 1989, Genetic algorithms in search, 

optimization, and machine learning, Addison Wesley. 

[5] Shahookar, K. and Mazumder, P., 1990, A genetic 

approach to standard cell placement using meta- 

100 

genetic parameter optimization, IEEE Trans actions 

on CAD, 9/2: 500-511. 

[6] Ma zumder, P. and Rudnick, E. M., 1999, Genetic 

algorithms for VLSI design, layout & test automation, 

Prentice Hall PTR. 

[7] Rahmani, A.T. and Ono, N., 1993, A genetic 

algorithm for channel routing problem, Proceeding of 

the 5th International. Conference on Genetic 

Algorithms, 494-498. 

[8] Goni, B. M., Arslan, T. and Turton, B., 2000, A 

genetic algorithm for over-the-cell and channel area 

optimization, Proceeding of the 2000 Congress on 

Evolutionary Computation 1:586-592. 

[9] Xilinx Inc., 1997, XC4000E and XC4000X series 

FPGAs, Data sheet. 

[10] Rose J., Gamal, A. E. and Sangiovanni-vincentelli, 

A., 1993, Architecture of Field-Programmable Gate 

Arrays, Proceedings of the IEEE, 81/7:1013-1029. 

[11] Brown, S., Francis, R. J., Rose, J. and Vranesic, 

Z.G., 1992, Field Programmable Gate Arrays, Kluwer 

Academic Publishers . 

[12] Betz, V., Rose, J. and Marquardt, A., 1999, 

Architecture and CAD for Deep-Submicron FPGAs, 

Kluwer Academic Publishers . 

[13] Baker, J. E., 1987, Reducing bias and inefficiency in 

the selection algorithm, Proceedings of the 2nd 

international conference on genetic algorithms and 

their applications, 14-21.

MAINTAINING ELECTROMAGNETIC COMPATIBILITY THROUGH 

DESIGN FOR FABRICATION AND ASSEMBLY OF PRINTED CIRCUIT 

BOARDS (PCBs) 

Tom Page, * Paul Baguley and * Dirk Schaefer 

Department of Design & Technology, Loughborough University, Loughborough, Leics. UK. 

* School of Engineering, Durham University, Durham, UK. 

ABSTRACT 

Recent regulations have demanded that electronics manufacturing companies control emissions 

from their products and the susceptibility of their products to emissions from other products. In 

addition, unexpected product failure and the ever-present demands of technology are also forcing the 

electronics industry to face the need to maintain electrical integrity. 

Our investigations into high-speed design techniques have shown three major causes of failure: 

emissions from interconnecting conductors; poor PCB layout and lack of technical knowledge in 

electromagnetic compatibility (EMC). Catching these kinds of electrical integrity problems early in the 

design phase allows designers to take timely action without jeopardising project time scales. The 

work reported here presents design for manufacturing guidelines and rules to maintain electrical 

integrity in printed circuit boards (PCBs). Currently, a common method for handling EMC is through 

compliance testing of the final product. Similarly, noise budget is measured on finished prototypes. 

Since product life cycles are reducing, dealing with EMC late in the design cycle is undesirable. The 

cost of fixing may also be higher at a final stage because only a few options are available to correct 

the problem. A ‘find and fix’ approach is no longer acceptable anymore. 

More and more companies are facing or will soon be facing EMC and electrical integrity issues. 

The majority of analysis tools available today are targeted toward simulation engineers. Such tools 

are not easy to use and are dependent on the availability and accuracy of complex simulation models. 

Moreover, they also tend to be ineffective on how to correct potential EMC problems. 

KEYWORDS: Design for Manufacture, EMC, PCB Design 


The electronics industry is ever-changing, a 

good indication of these changes is to consider 

microprocessor performance over the last few years. 

The graph shown in Figure 1 illustrates the past, 

present and future trends in microprocessor clock 

frequency thus necessitating the need for signal 

integrity issues so as to prevent or minimise the effects 

of electro-magnetic interference. 

Interconnecting traces exhibit “parasitics” such 

as self inductance, self capacitance, mutual inductance 

and capacitance with their surroundings. These 

parasitics are very small factors with signals running at 

low frequencies, but cannot be ignored at high speed. 

Presently, the continual increase of clock frequencies 

and faster rise and fall times, cause interconnects to 

start act active as active elements of the design and 

interconnect delays start to become even greater then 

component delays. In other words, at higher 

frequencies, a signal trace becomes an interconnect 

component. A PCB trace alone can be represented as 

101 

a transmission line consisting of series inductance and 

shunt capacitance elements distributed along the line. 

At high speed, a signal propagating down the line 

has to charge up each inductor and capacitor element 

before it is passed to the next element. It is intuitively easy 

to understand that this charging process has the effect of 

reducing the propagation time and increase the impedance 

of the trace. 

The characteristic impedance (electromagnetic wave 

resistance) of a PCB trace is determined primarily by its 

width, the PCB material and the thickness between signal 

layer and power plane. The physical construction of the 

PCB may be considered as a wave-guide made up of trace 

geometry, dielectric insulator and planes. A typical PCB 

layer cross-section can be can be divided in five different 

configurations: microstrip; embedded microstrip; dual 

microstrip; stripline and dual stripline. Each configuration 

has its own characteristic impedance and propagation 

delay.

MHz 

100,000 

10,000 

1000 

100 

10 

1 

0.1 

Impedance is expressed as: Z = 

80286 

8086 

1486 TM 

Processor 

1386 TM 

Processor 

Pentium- Pro® 

Processor 

Pentium® 

Processor 

‘75 ‘80 ‘85 ‘90 ‘95 ‘00 ‘05 ‘10 ‘15 

Projected 

10GHz 

Figure 1:Moore’s Law, Frequency with respect to time 

L ---------- (1) 

C 

2 CROSS-SECTIONS OF VARIOUS 

TRANSMISSION LINE CONFIGURATIONS 

Microstrip : A track routed over a solid ground plane. 

Embedded Microstrip: Same as microstrip but trace is 

buried in the insulator. 

Dual microstrip: Two surface traces buried in the 

insulator. 

Stripline: Case of a wire sandwiched between two 

planes. 

Dual stripline: Two traces centred between two planes. 

With reference to figure 2, a signal runs faster on 

microstrip configuration than on stripline. So for a given 

delay, longer tracks are permitted with a microstrip 

configuration. It is important to note that, a wire routed 

in microstrip is not shielded from emissions by the 

Microstrip Stripline 

Embedded Microstrip Dual Stripline 

Dual Microstrip 

Source: Intel 

L being the per unit length 

inductance and C the per unit 

length capacitance. 

power and ground planes as it would be the case of a wire 

routed in stripline. 

Shielding in case of stripline gives typically a 

reduction of emissions by up to 10bB. For high speed 

boards, place power and ground planes directly adjacent. 

This will maximise the capacitive coupling and thus reduce 

supply noise. Also use extra ground planes (and not pwer 

planes) to isolate routing layers. For example, for an 8 

layer PCB (4 routing), the best assignment for EMC 

performance is S1, G, S2, G, P, S3, G, S4 where S = signal 

routing layer, G = ground plane and P = power plane. 

Figure 2: Cross-sections of various transmission line configurations 

102

3 PROPAGATION DELAY 

Propagation Delay is a complex function of many 

parameters including incluing inpedance and loading. 

Intended functionality requires that timing constraints 

have to be defined. Here are two examples. Delay 

control, see figure 3, if Tpd is too long, then the pulse will 

arrive too late for the intended function. 

Figure 3: Delay control 

Skew management, see figure 4, the path delay of 

A+C OR B+C must be equal to D within device tolerance. 

Important factors for skew management are gate 

propagation delays and copper trace delays. Effective 

skew management is ideally performed in the time 

domain not in the geometrical domain. 

Figure 4: Skew management 

4 ATTENUATION 

At high frequencies, current density no longer 

becomes evenly distributed in the cross-section of the 

trace. The current density becomes higher near the 

surface of the trace and the resistor of the conductor 

increases. This skin effect has a consequence on the 

high harmonic frequencies and is responsible for signal 

degradation and attenuation. Because of the skin effect, 

a square pulse shape will become slightly rounded. Also, 

while the resistance increases with the frequency, the 

intrinsic trace inductance reduces and it is then the loop 

inductance that becomes the dominant paramater in 

thecharacteristic impedance. This inductance is a 

function of the geometry of the signal and its ground 

return path. In order to minimise these effects with high 

speed signals, short and wide tracks are highly 

recommended. 

103 

Original 

Effects of attenuation 

Figure 5: Effects Attenuation on an original pulse 

5 REFLECTIONS 

A signal propagating down a lossless line of 

constant characteristic impedance will travel along the 

line without distortion. However, when the signal arrives 

at the end of the line (after Td time), reflection will occur if 

the load impedance does not match the impedance of the 

conductor, see figure 6. The reflected wave travels back 

down the line and after Td time, it hits the source and is 

reflected back again but now from the source. 

Figure 6: Chronlogy of a reflection 

The signal is partially reflected, the amount of 

reflection is dependent on the magnitude of the 

impedance mismatch. The reflection coefficient, 

expressed as a percentage can be calculated as: 

Kref 

( Zs − Z0) 

100× 

( Zs + Z0) 

= ------------- (2) 

When the rise time of the signal is greater than the 

propagation delay down the trace, the reflections are 

masked by the slow rise of the signal. If the two-way 

propagation delay (source-end-source) is longer than the 

rise or fall times, then the reflection from the far end 

arrived after the initial transition is finished. The length 

beyond which the line should be terminated is given by: 

Tr 

L = 

2× 

Td 

------------- (3)

Where: L = length of trace, Tr = edge rate, Td = 

loaded propagation delay. 

Long nets may be subject to high amounts of 

reflection. For example, rise times of 2 nano-seconds 

require special consideration when the trace length is 

greater then 14 cm (for the stripline topology, FR-4 

material). This length threshold is often called the critical 

length, it is the maximum permissible unterminated trace 

length. Different logic families with characteristic rise 

times will demand critical length of traces in order to 

reduce the impedance. Ensuring that traces are kept 

under their critical length is often unpractical. In such 

cases a termination scheme may be the best solution. 

6 TERMINATION STRATEGIES 

As mentioned earlier, in order to reduce 

reflections, the input impedance of the receiver must 

appear to match the impedance of the inductor. The 

reflection coefficient, expressed in percentage is 

calculated from equation (2). An input impedance of the 

receiver, larger than the conductor’s impedance will 

cause over/undershoot, while the opposite will cause a 

drop in the pulse. The side affects of termination area: 

additional of extra delay to the signal; extra power 

consumption due to current requirement; and extra cost of 

component. Typically, such termination methods are 

recommended by the semiconductor vendor. Methods of 

termination comprise the following: series resistor; parallel 

resistor; Thevenin (split resistor); RC termination and 

diode clamp. With the exception of the diode clamp 

strategy, all other termination methods involve the output 

resistance of the driver that is non-linear and varies for 

most semiconductor processes over a wide range. 

7 TOPOLOGY 

Timing constraints are taken into account when 

selecting the appropriate net topology to control the 

arrival time of the signals at their respective receivers. 

Various topologies are used to achieve different 

objectives, figure 7 illustrates commonly used net 

topologies. 

Figure 7: Various net topologies 

The star topology is characterised by multiple 

branches from a central driving point that requires a 

strong driver. In this topology, series termination can be 

used, on each leg. However, if branch length matching is 

not possible, then RC termination close to central point 

can be used. In the case of the daisychain topology 

where loads are distributed down the line, together with a 

strong driver there will be skew between loads. As such, 

parallel, Thevenin or RC termination can be used. In 

point-to-point, connection between two points, when 

104 

termination is required, a series resistor at the source is 

recommended, with a resistor value optimised for a rising 

or failing edge. This is dependent on how balanced the 

high/low impedance of the driver. In the case of TEE (Far 

end cluster), essentially a variation of star topology, 

where loads are close together and skew must be 

minimised. In this topology it is possible to use either 

series termination close to the driver, or RC termination at 

the branching point to the loads. Finally, the H tree 

topology is easier to drive than a star topology. With this 

method, each line from a tap to a load has the same 

impedance, so the same termination resistor value can be 

used at each load. The other advantage of the H-tree 

structure is that the wire length between the driver and 

each load is identical, preventing such signal skew 

problems. 

8 GROUND BOUNCE 

Outputs are inductively coupled between power 

and ground, as the component switches state. The 

sudden high current requirement will lead to a reverse 

voltage drop called ground bounce or switching noise. To 

minimise the effect, decoupling capacitors should be used 

(with low lead self inductance) and placed close to the 

component to be decoupled. The capacitor maintains a 

constant voltage across the component and delivers a 

more stable current outside of the general power planes. 

Since ground is more sensitive than VCC, the decoupling 

capacitor should be located as close as possible to the 

ground pin. Extra capacitance is provided by the use of 

multilayer PCB power and ground planes. In fact, ground 

bounce was the main contributing effect to the change 

from double sided to multilayer PCB with power and 

ground planes. Care should be taken when planes are 

getting heavily perforated, because this will cause the 

plane inductance to increase. Currents in an imperfect 

ground plane flow from ground pins to the power 

connector and can affect the voltage of other components 

with ground pins. High ground currents in the plane are 

more likely to occur with bus and high current drivers, 

they should therefore be located close to the power 

connector. It is important to ensure that self-resonant 

frequency of decoupling capacitors is above the 

frequency of the signal to be decoupled. 

9 CROSSTALK 

A high speed trace behaves like a transmitting 

antenna. A sudden change of current on one trace can 

cause capacitive and inductive coupling into adjacent 

traces. When this crosstalk level is sufficient, a false 

signal transition can occur. The amount of coupling 

between traces is proportional to the length of the two 

traces running in parallel and inversely proportional to 

their spacing. Strategies for the reduction of crosstalk 

include: minimising the length of parallelism; maximising 

trace spreading; lowering the thickness of the dielectric; 

minimising impedance; minimising the edge rate and 

selecting the most appropriate termination strategy. 

10 CONCLUSIONS EMI REDUCTION TECHNIQUES 

In order to reduce electromagnetic interference, 

there are a number of approaches including: minimising 

rise and fall times on clock and signal edges (see figure 

8); utilising slowest logic consistent with the circuit 

operation (see figure 9); and board partitioning to 

minimise ground bounce effects (see figure 10).

Figure 8: Minimising rise and small times on signal and 

clock edges 

Minimised rise and fall times on signal and clock 

edges. Sharper edges cause high harmonic contents in 

the high frequency spectrum. 

Figure 9: Using Slowest logic consistent with the circuit 

operation 

Selection of the most appropriate logic family plays 

a key role in the overall EMC performance. Logic families 

that are designed to operate at high frequency will exhibit 

sharp edge rates. This will result in larger harmonic 

spectral contents. Figure 9 is ordered by the noisiest at 

the top to the less noisy at the bottom. The ECL family is 

relatively less noisy, this is simply because it has the 

small voltage swing. CMOS is slow and also has reduced 

drive capability, this is why it is the best from an EMC 

standpoint. 

Figure 10: Board partitioning 

To minimise ground bounce effect, it is desirable 

to place high speed components close to the power 

source with slower components placed further away. To 

minimise cross coupling and thus system noise it is also 

advisable to create separate partitions for analogue and 

105 

digital sections. Furthermore, to prevent noise conducted 

through input/output IO cables, segregate IO connectors, 

IO drivers/receivers and non-IO components. Provide 

extra spacing between highly radiating nets and IO nets. 

11 REFERENCES 

1. Anzivino, Rich (2002): The DFA Design 

Philosophy, Printed Circuit Design, pp.14-26. 

2. Atiyeh, Phillip G. (2003): Design for Assembly: 

Sometimes More is Less, Assembly Automation 

Vol. 12 No. 2, pp.26-40. 

3 Classon, F. (2003): Component Placement, 

Surface Mount Technology Vol. 8, 4. pp.45-62.

ENSURING ELECTROMAGNETIC COMPLIANCE IN PRINTED CIRCUIT 

BOARDS THROUGH DESIGN FOR ASSEMBLY GUIDELINES 

ABSTRACT 

Tom Page, *Paul Baguley and *Dirk Schaefer 

Department of Design & Technology, Loughborough University, UK. 

* School of Engineering, Durham University, Durham. UK 

The aim of this work is to provide the embodiment of sample PCB design for manufacturing 

and assembly rules for fine-pitch devices in a commercially PCB design application. Most 

commercially used PCB design applications provide for PCB design rule-checking to an extent, 

particularly in setting geometric constraints for laying out components on the substrate. For 

example, in automatic placement of connectors and components within a PCB design file. The role 

of the design rule checker within PCB design software is a fundamentally important in ensuring 

process yield in PCB fabrication as well as ensuring functional acceptance in test whilst maintaining 

accurate placement of components. 

The design rules implemented in this paper emanate from a wider programme of research 

that focusses on design for manufacturing and assembly issues in Surface Mount Technology 

(SMT), Fine-Pitch Technology (FPT) and Flat Pack Ball-Grid Array (FPBGA). This work explores 

the methodology of decision support in the design of electronic products with specific regard to 

design for manufacturing and assembly. The design for manufacturing rules used in this 

implementation scenario is for the precise location and placement of Fine-pitch components. 

1 DESIGN REQUIREMENTS FOR ULTRA FINE- 

PITCH DEVICES 

In order to control overall area requirements for the 

higher pin-count devices, component manufacturers have 

reduced the lead pitch (the space between lead centres). 

The standard family of the Standard Quad Flat-Pack 

(SQFP) has been adopted by the industry to provide 

commercial packaging of custom and semi-custom 

integrated circuits with 0.65mm, 0.5mm, 0.4mm and 

0.3mm lead pitch. Although assembly processing has 

been refined to acceptable yields levels for 0.65mm and 

0.5mm pitch device types, conventional process methods 

might not be practical for high pin count using the 0.4mm 

and 0.3mm pitch devices [1]. While adapting this 

advanced packaging technology, the users of the SQFP 

are challenged by several considerations, including 

physical, financial and environmental issues. Physical 

issues include attachment processes and finished product 

reliability. Figure 1 illustrates the design guidelines and 

rules that require to be considered and adhered to in the 

design for fabrication and assembly of fine-pitch PCBs. 

106 

Specific reference is made to abbreviations such as 

CSG1…CSG10 these correspond to discrete fine-pitch 

design rules which have been coded into a knowledgebased 

decision-support tool. This was the main 

deliverable from the research programme from which 

Figure 1 was taken. 

1.1 Designing the Substrate for Fine-Pitch 

Assembly 

Many factors have impact upon assembly yield as 

well as solder joint quality. Each of the following 

elements must be carefully reviewed and calculated so as 

to ensure ease of assembly. Formulation the land pattern 

array and calculate geometry such that devices are 

placed exactly on the solder lands. Implementation of the 

test model substrate design layout so as to ensure that 

the fabrication detail is to specification prior to assembly. 

The small SQFP devices are manufactured in high 

volume and provide rugged, long-term reliability. Newer 

packages in the industry, such as the 0.4mm (0.016-in) 

lead pitch SQFP might pose a challenge to manufacturers 

in maintaining a consistent assembly process yield [1].

Component Selection 

Guidelines: 

CSG1...CSG10 

Verification of Post 

Assembly Evaluation : 

PAE1...PAE14 

Verification of PCB 

Fabrication Details: 

VFD1...VFD10 

Verification of Assembly 

Documentation: 

VAD1...VAD16. 

Verification of Bare 

Board Dimensional 

Accuracy: 

VBB1...VBB16 

Land Pattern Design 

Guidelines: 

LPD1...LPD11 

Design for Test 

Rules: 

DTR1...DTR6 

Design 

Requirements for 

Fine-Pitch Device 

Fabrication & 

Assembly 

Fine-Pitch 

Design Rules: 

FPR1...FPR17; 

FSC1...FSC11. 

PCB Layout Design Guidelines & Rules: 

GLD1...GLD18; DLD1...DLD16; 

GDR1...GDR12; DLR1...DLR13. 

PCB Routing Design 

Guidelines & Rules: 

RDG1...RDG15; 

DRD1...DRD18. 

Solder Control 

Design Rules: 

SCR1...SCR16. 

Mixed Technology 

PCB Design Rules: 

MTD1...MTD18. 

Axial-Leaded & Plated 

Through Hole Design Rules: 

ADR1...ADR16. 

Figure 1: A Model of Knowledge-Based Decision Support in PCB Design for Assembly 

1.2 Typical Fine-Pitch Device Package Description 

The EIAJ standard SQFP 256-pin configuration on 

0.4mm (0.016-in) pitch provides adequate termination 

channels for more advanced ASIC devices while requiring 

a relatively small area for attachment. The total surface 

area (lead end to end) of the 256-lead plastic device is 

less than 31.0mm (1.22-in) square with a maximum 

height from the substrate surface of 3.5mm (0.137-in). 

The leads are formed in the gull-wing configuration 

extending from the device body providing a contact area 

of 0.5mm±.0.2mm (0.02-in±0.008-in) [1]. For a process 

development or test program, a non-functional component 

can be ordered with internal bonding on lead pairs before 

moulding. 

1.3 Design Allowance for Physical Tolerances of 

Devices 

Although the lead-pitch tolerance can be defined 

as non-accumulative, the width of the lead at the 

attachment area is specified as wide as ±0.07mm (0.003in). 

Using the maximum material condition of the basic 

lead width of 0.15mm opens the possibility for up to 

0.22mm total lead width. The final land pattern geometry 

provided on the substrate must accommodate this 

maximum material condition of the device so as to avoid 

the possibility of lead overlap at either the toe or heel of 

the J-lead termination. The land pattern geometry and 

spacing between pad rows are derived from calculating 

both the maximum and minimum material conditions of 

the device. That is, if the device is supplied at the 

maximum overall width (lead end to lead end), the land 

pattern should provide enough surface area to prevent 

lead overhang [1]. 

107 

1.4 The Impact that Design Has on Assembly 

Efficiency of Fine-Pitch SMT PCBs 

Seventy percent of the failures detected on surface 

mount assemblies are due to solder defects. Solder 

defects found at test are either in the form of a short 

between device leads or open circuits due to insufficient 

solder [2]. An ongoing process audit can reduce most of 

the solder defects but often manufacturing problems are 

design related and if not corrected will remain a source of 

chronic failure. By the implementation of proven circuit 

board design rules it is possible further reduce solder 

defects. These rules specifically employ processcompatible 

land pattern geometries for surface mount 

devices and each will play a significant role in ensuring 

manufacturing efficiency and quality. The design rules 

will also focus on PCB fabrication guidelines and 

assembly machine compatibility. Land pattern geometry 

on the other hand is directly related to process control 

and solder attachment uniformity. When each of these 

primary disciplines are properly defined and implemented 

all solder defects can be eliminated. Solder defects can 

be further reduced through continued process refinement. 

Each process step must be monitored by means of 

human or automated visual inspection during the initial 

start-up of a product, 100% inspection is not uncommon. 

After process stabilisation, only sampling of the assembly 

is needed. As each unit is inspected, defects exceeding 

standard limits are identified and recorded. The defect 

ratio from one assembly to the other does not seem to be 

affected by the reflow process as much as the solder 

paste characteristics [3]. The viscosity of the solder paste 

will have a more significant impact on the solder joint 

defect ratio of the 0.4-0.5mm (0.016-0.020-in) devices. 

Although the solder volume of each joint can be 

mathematically modelled, visual inspection on the finer 

pitch devices is not without compromise. A 100%

inspection of every lead on the finer pitch devices is not 

practical and the use of advanced inspection systems for 

fine-pitch devices is inevitable. For example, ultrasonic 

imaging, X-ray and X-ray laminography might prove to be 

very effective in performing a non-destructive solder 

quality certification to measure solder density of detection 

of solder voids [4]. Solder paste registration for the 

majority of the surface mount devices can be improved, 

for example, by reducing the overall stencil opening and 

to compensate for all the tolerance variables of a typical 

PCB, a reduction of the stencil opening by 10-20% might 

be adequate. 

1.5 Placement Accuracy and Land Pattern Design 

Requirements 

Less-than-perfect machine placement of passive 

and 1.27mm (0.050-in) pitch surface mount devices can 

be tolerated to a limited extent. As during the reflowsoldering 

process the entire assembly is heated to 

approximately 200°c and the solder paste is converted to 

a liquid state. These devices are momentarily suspended 

in the liquid alloy and through surface tension, the device 

mass will tend to self-centre before the solder begins to 

cool. This phenomenon while predictable for passive and 

coarse-pitch devices, cannot be relied upon for fine-pitch 

attachment as the land pattern geometry is much more 

critical. The land pattern geometry must provide for both 

the minimum and maximum material condition of the 

device and substrate [5]. The designer must calculate the 

basic land pattern limits with provision for inspection and 

machine placement tolerances. 

2 USING PCB DESIGN RULES WITHIN PCB 

DESIGN SOFTWARE 

With PCB design software the PCB is designed by 

the placement of components, tracks, vias and other 

design objects. These objects must be placed in the 

workspace with close regard to each other. Components 

must not overlap, nets must not short, power nets must 

be kept clear of signal nets, etc. To allow the designer to 

remain focused on the task of designing the board, a 

design rule checker can monitor such design 

requirements. These design rules are monitored as the 

designer lays out the PCB. As soon as an object is 

placed in violation of a design rule it is highlighted. 

obeyed. 

3 AN IMPLEMENTATION OF PCB ROUTING 

DESIGN GUIDELINES FOR EMC 

A PCB was selected for the implementation of 

such design rules it was a digital-to-analogue converter 

board used in a radio frequency product. The design of 

the PCB was constrained by the product envelope and 

therefore had to very small in size, specifically 40mm by 

75mm. The fine-pitch devices used in this design had 

lead-pitch of 0.4mm. Owing to the fact that the product 

was a radio frequency application, specific attention was 

made to design for Electro-Magnetic Compatibility (EMC). 

EMC is characterised by capacitive and inductive 

reactions within signal traces because such traces can 

act as unintentional wave-guides for the reception and 

transmission of radio frequency signals. These reactions 

are often referred to as electrical noise and PCB 

designers are required to exercise caution in laying out 

components and routing traces on PCBs so as to 

minimise such noise. 

To ensure that the PCB was EMC compliant 

specific design rules pertaining to signal trace-widths and 

corresponding air-gaps between traces need to be 

adhered to. As a rule of thumb for EMC, the design-rule 

checker needs to be instructed that signal traces cannot 

be less than 0.13mm wide and the gap between signal 

traces must not exceed 0.13mm. Moreover, for power 

and ground planes namely VCC and GND the respective 

trace-width cannot be less that 0.2mm and the gap 

between such traces must not exceed 0.2mm. With 

reference to PCB Routing Design Guidelines it is possible 

to implement such rules within the routing design rule 

class, under width constraint and clearance constraint 

(see Table 1). Table 1 shows the definitions of the design 

rules for width constraint and clearance constraint, it 

explains what happens when a design-rules violation 

occurs and explains when the design rule can be invoked 

by the designer. 

The implementation of PCB routing rules Figures 1 

and 2 respectively. Figures 1 and 2 are design-rule entry 

dialogs for 'Routing Width Constraint' and 'Clearance 

Constraint' design rules respectively. 

Routing Width Constraint design rule Clearance Constraint design rule 

Rule class: Routing Rule class: Routing 

Defines the minimum and maximum width of Defines the minimum clearance allowed between any 

tracks and arcs on the copper layers. 

two primitive objects on a copper layer. Use the 

Clearance Constraint to ensure that routing 

clearances are maintained. The Connective 

How Duplicate Rule Contentions are 

Checking option would typically be set to Different 

Nets. An example of when Any Net could be used is 

to test for vias being placed too close to pads or other 

vias on the same net, or any other net. 

How Duplicate Rule Contentions are Resolved: 

The rule with the largest clearance is obeyed. 

Resolved: The rule with the tightest range is 

Rule Application: During auto-routing and 

Batch DRC. 

Rule Application: On-line DRC, Batch DRC and 

during auto-routing. 

Table 1: Routing Width Constraint and Routing Clearance Constraint Design Rules. 

108

Figure 1: Rule Implementation in PCB Routing Width Constraint Rule-Check Dialog 

Figure 2: Rule Implementation in PCB Clearance Constraint Rule-Check Dialog 

109

4 CONCLUSION 

This paper demonstrated the implementation of 

sample fine-pitch PCB design rules in an existing PCB 

design file. The design rules implemented were a sample 

of those guidelines and rules represented selected from a 

programme of wider study in design for assembly in PCB 

design. This sample was selected on the basis that these 

design rules lent themselves to quantitative and 

geometric representation. Those PCB design rules that 

were implemented included: routing design guidelines; 

layout design rules; and device oriented layout rules. The 

PCB design-rule editor is instructed of such design 

requirements by the setting up of a series of design rules. 

Also, during the board verification process, an integrated 

design rule checker can be executed which will generate 

a report of any design rule violations on the PCB. 

5 REFERENCES 

1. Chroneos R.J. et. al.,(June 1996): Packaging 

Alternatives for High Lead Count, Fine Pitch, 

Surface Mount Technology, IEEE Transactions on 

Hybrids and Manufacturing Technology, Vol.16, 

No.4, pp.396-401. 

110 

2. Li Y., Mahajan R.L., Tong J., (June 1998): Design 

Factors and Their Effect on PCB Assembly 

Yield, IEEE Transactions on Components, 

Packaging, and Manufacturing Technology, 

Part A, Vol.17, No.2, pp.183-191. 

3. Neger V., Pawlischek H., (2000): Integration and 

Application of Vision Systems in SMD Automatic 

Placement Machines, Circuit World, Vol.17, No.1, 

pp.24-29. 

4. Stevens M., Ball E., Protogeros A., (2002): 

Process Compensation and Printed Circuit Board 

Manufacture, International Journal of Advanced 

Manufacturing Technology, No 8, pp.85-90. 

5. Wassink R.J.K., (2001): Footprints of Fine Pitch 

SMDs, Working Paper Nederlands Philips 

Bedrijven BV Centre For Manufacturing 

Technology.

Knowledge Transfer in Networks of Competence 

Prof. Dr.-Ing. habil. Christian-Andreas Schumann, University of Applied Sciences Zwickau / VDI Representation Saxony 

Dipl.-Inf. Claudia Tittmann, University of Applied Sciences Zwickau 

Dipl.-Inf. Kay Grebenstein, Middle German Academy of Further Education, Zwickau 

Abstract 

Networks of knowledge transfer have a lasting effect on the parallel existing networks of competence. An essential problem is 

the development of content modules from individual and objective knowledge. The potential of regional networks of 

knowledge transfer for solving this problem will be introduced. These networks are based on special projects concerning the 

regional technology transfer. The activities are focused to the computer based engineering intelligence. 

Keywords: 

Knowledge Transfer, Engineering Intelligence, Competence Cell 

1 NETWORKS AND CELLS OF COMPETENCE 

The recent development in Knowledge Transfer is only to 

understand in the context of national and international, 

interdisciplinary corporation of public, non-profit and 

business units. The expenditures of the required 

resources to determine the subjective knowledge, to 

transfer it into evaluated know how of the organisation, 

and to include it into transfer programs is so enormous 

that there is a strong constraint for organisations to 

cooperate in tailored networks of specialists, developers 

and users (Figure 1). 

Figure 1: Knowledge Flow. 

These networks are well known as networks of 

competence based on so called competence cells (Figure 

2). Each cell is a separate organisational unit 

characterised by independence in law and economy, 

special knowledge, skills, performance, competence and 

interfaces. 

111 

Figure 2: Competence Cell. 

The separate cells form the temporary or constant 

networks of competence (Figure 3) in the framework of 

business, engineering and production activities and 

processes in order to be successful in the market and 

product development. The theory of networks of 

competence was developed especially for the production 

networking, but it includes the networking of 

interdisciplinary knowledge transfer. 

Figure 3: Competence Cell in the Network.

Corporation and networking have to be managed. 

Besides the general management and project 

management especially the management of data, 

information, skills as well as knowledge is essential 

(Figure 4). 

Figure 4: Context of Competence, Skill, Knowledge. 

The common information management is supplemented 

by the knowledge management. Knowledge is the ability 

to define patterns of facts serving the preparation and the 

execution of actions and decisions. The similar relation is 

remarkable in the case of the general business with 

regard to the management of e-based processes and 

functions. 

All sides of management, business, production, and 

engineering including the special fields of information 

management and knowledge management as well as the 

e-business, the computer integrated manufacturing up to 

the e-learning are influenced by and related to the 

business, engineering, and production intelligence. 

Figure 5: Importance of Knowledge Transfer. 

Business, engineering and production intelligence is the 

ability to generate knowledge by analysing the state of the 

art in the own as well as in the competitive business, 

engineering or production units and networks (Figure 5). 

The competence cells for knowledge transfer are 

arranged in chains consisting of development, supply, 

and application modules. Due to the scale of the 

competence of the organisational unit the cells are single 

or multi functional. The multi functional cells include 

integrated functions and processes of development, 

supply and/or application. According to the integrated 

functions and processes the competence cells are put in 

its proper place in relation to the other cells forming the 

network of competence in all. The networking is one of 

the main aspects guaranteeing the success of the 

business, engineering, production or learning unit, 

respectively. 

112 

2 KNOWLEDGE TRANSFER IN FRAMEWORK 

PROJECTS 

The key question is to prepare the individual knowledge 

for generating content modules with objective knowledge 

transferred in or outside of the organisation. Therefore as 

an example the network of regional knowledge transfer 

was founded. 

Figure 6: Real Competence Network. 

The representatives of the competence cells discussed 

and developed the transfer concept with special 

technology-oriented modules mainly focused for 

supporting SME. The modules are designed to constitute 

a construction set for content modules for knowledge 

transfer from the technology leaders to the other 

enterprises (Figure 7). Each competence cell generates 

its special subprojects due to the special profile of the 

organisation. 

Figure 7: Model of Sections. 

The knowledge of the modules is connected by defined 

interfaces. The controlling activities were done by the 

strategic partner of the network of competence: the 

society of German engineers (VDI). The experts of the 

VDI evaluate the detailed proposals for each module in 

order to check the level of knowledge and the relation to 

other contents (Figure 8). After the successful evaluation 

the knowledge transfer module is developed by the 

competence cell and offered the consumer contractors of 

the network of competence or the other interested 

partners.

Figure 8: Modules. 

For the knowledge transfer the recent technology is used. 

This means new kinds of distance education and blended 

learning, tele-teaching and tele-working are included. For 

instance new WBT-programs were developed for 

Enterprise Resource Planning, Business Intelligence, 

Business Process Reengineering and Optimisation, 

Media Competence, etc. used by the access via internet 

portal, tele-platforms and learning spaces (Figure 9). 

Figure 9: Learning Platform. 

The programs are embedded in courses for students and 

employees in the framework of tutoring and blended 

learning (Figure 10). Some of the recent offered or 

planned knowledge transfer modules are innovative plant 

planning, business process reengineering, and total 

quality management as well as business intelligence 

engineering, and flexible automated assembly systems. 

Figure 10: Map of Page. 

For the complexity of the task a workflow system is 

113 

essential, and some special systems solutions were 

developed by the competence cell itself especially in 

order to improve the performance of the whole system 

(Figure 11). 

Figure 11: Quality Assurance. 

3 REGIONAL TECHNOLOGY TRANSFER IN SPECIAL 

PROJECTS 

The general framework system for knowledge transfer is 

based on special projects concerning the regional 

technology transfer. The activities are focused to the 

intelligent aspects of the industrial development: the 

computer based engineering intelligence (Figure 12). 

Figure 12: Modules with focus on engineering intelligence. 

One of the most important factors is the innovation in 

planning and realisation of factories. Therefore one recent 

subproject deals with the innovative concepts for factory 

planning. The main topics for the subproject are: 

1. Basics and Introduction such as evolution of the 

industrial production, recent trends and requirements, 

visions and strategies 

2. Factory of the Future such as production in networks, 

modular flexible factories (Figure 13)

Figure 13: SCM-Systems view. 

3. Innovative Concepts such as holistic factory, bionic 

manufacturing, digital factory 

4. Process-oriented Analyses and Design such as event 

driven process chains, workflow analyses and design, 

data process analyses and design 

5. Manufacturing and Control Techniques such as 

enterprise resource planning, just in time, control systems 

and stations 

6. Maintenance Management such as preventing 

maintenance, special maintenance technologies 

7. Human Resources such as motivation of employees, 

self control regulation, self and external controlling of 

performance units 

8. Information Management and Technology such as 

information management, product data management, 

knowledge management, computer integrated 

manufacturing, computer based plant planning, factory 

information systems, modelling and simulation (Figure 14) 

Figure 14: Solution for Information flows. 

9. Ecology and Factory such as environmental influence 

of production, environmental information systems 

10. Energy Supply for the Factory such as energy supply 

and transformation systems, optimisation by energy 

management. 

11. Practice of Application such as best practice or best in 

class solutions, excursion. 

Today each concept and solution is computer based. That 

is why the computer aided applications in engineering 

114 

were assimilated to each professional topic of the 

subproject. One of the central themes was the embedded 

computer integrated manufacturing in the modern factory 

design (Figure 15). 

Figure 15: Simulation view to a manufacturing process 

For instance the Computer Aided Design is important for 

the plant development as well as the facility management, 

the product development, the control and circuit design, 

the factory layout, etc. The existing knowledge of the 

professionals was presented in several special events by 

specialists supported by tutoring based on modern 

internet technologies of computer and web based training 

opportunities (Figure 16). 

Figure 16: Solution – KM-System. 

The result was an agreement of the partners to continue 

the knowledge transfer in existing networks of 

competence such as the association of the tool machine 

industry in Saxony and the association of the automation 

companies in the region. 

4 SUMMARY 

The knowledge transfer by the network of competence 

works successful and efficient. Therefore the extended 

corporation with other networks of competence is in 

progress. Dependent on the target group for the 

application of the knowledge transfer, the competence 

cells form modified or new networks of competence. 

There is a very flexible and economical way of knowledge 

transfer under construction and partly in application. The

knowledge transfer will be promoted by new methods, 

contents and software. The best practice applications are 

used to push the competence cells and their integrated 

networks. 

5 REFERENCES 

[1] Chr.-A. Schumann, Claudia Tittmann et. al., “AIbased 

integration of business intelligence and knowledge 

management in enterprises”, AIAI 2004 Toulouse, 2004. 

[2] Chr.-A. Schumann, Claudia Tittmann et. al., “Media 

Competences and skills for web-based Knowledge 

transfer and Blended learning”, EISTA 2004, Orlando, 

2004. 

[3] Chr.-A. Schumann et. al., “Tele-Education and 

Blended Learning in complex Networks of Competence”, 

Euromedia-APTEC 2004, Proceedings, 2004. 

[4] Claudia Tittmann, “Brokerage und Controlling der 

Kopplung von Kompetenzzellen zu 

unternehmensübergreifenden Workflows in 

Leistungsnetzwerken“, Exposé Dissertation, 2004. 

[5] Chr.-A. Schumann, Claudia Tittmann et. al., “Process 

Support and Quality in WBT Content Development”, 

EDEN Proceedings 2003, Rhodes, Greece, 2003. 

[6] Chr.-A. Schumann et. al., “Media Competence for the 

Management of Small and Medium Sized Enterprises 

(SME)”, EDMAN 2003 Proceedings, Brno, Ceska, 2003. 

[7] Chr.-A. Schumann, Claudia Tittmann et. al., „Globale 

IT-Infrastruktur für die interkulturelle Kommunikation“, 

Kommunikation in der globalen Wirtschaft, Peter Lang, 

Frankfurt Berlin Bern Bruxelles New York Oxford Wien, 

2003. 

[8] Chr.-A. Schumann et. al., “Weiterbildung 

Medienkompetenz für KMU der Region”, 16 th International 

Scientific Conference Mittweida (IKWM) Proceedings 

2003, 2003. 

[9] Chr.-A. Schumann et. al., „Regionaler Wissenstransfer 

(REWITRA)“, Learntec 2003 Proceedings, Karlsruhe, 

2003. 

[10] Chr.-A. Schumann, Claudia Tittmann et. al., “Impact 

of the reorganisation of the enterprise information system 

115 

by knowledge management methods and tools on supply 

chains”, IBEC Proceedings Paris 2002, 2002. 

[11] Chr.-A. Schumann, “Knowledge Management”, 

Riemann: Wirtschaftsinformatik, Oldenbourg, 2001. 

[12] Chr.-A. Schumann et. al., “Improvement of Supply 

Chain Management by Knowledge Management in the 

Automotive Industry“, ATTCE 2001 Proceedings Volume 

3, Barcelona, 2001. 

[13] Chr.-A. Schumann, “Closing the Gap between School, 

university, and Further Education Systems in the process 

of life long learning”, EDEN – Open Classroom 

Conference, Proceedings 2000, Barcelona, Spain, 2000. 

[14] Chr.-A. Schumann et. al., “Humans and their Images 

in the Real and Virtual World of Learning, Training, and 

Working”, The Wanderstudent 2000. Proceedings, 

International Colloqium, KU Leuven, Belgium, 2000. 

[15] Chr.-A. Schumann et. al., “Development and Using of 

Regional Networks for Open Learning and Distance 

Education”, 19th ICDE World Conference on Open 

Learning and Distance Education Proceedings 1999, 

Vienna, Austria 1999. 

[16] Chr.-A. Schumann et al., “Hypermedia Based 

Distance Learning Systems and its Applications”, The 8th 

International Conference Information Systems 

Development ISD´99, Proceedings, Boise, Idaho, USA, 

1999. 

[17] Chr.-A. Schumann et. al., “Distance Education as 

part of modular-design system for lifelong learning”, XV. 

IFIP World Computer Congress Proceedings, 

Vienna/Budapest, Austria/Hungary, 1998. 

[18] U. Bentlage, P. Glotz, and I. Hamm, E-Learning 

Märkte, Geschäftsmodelle, Perspektiven, Verlag 

Bertelsmann Stiftung, Güthersloh, 2002. 

[19] B. Hall and J. LeCavalier, E-Learning Across the 

Enterprise – The Benchmarking Study of Best Practices, 

Brandon-Hall.com, 2001. 

[20] R. Sanchez, Knowledge Management and 

Organisational Competence , Oxford University Press, 

2003.

PART 6 

LATE PAPERS 

116

Fuzzy Sliding Mode Control Solution to an Inverted Pendulum : An 

Implementation Issues 

Mouloud Bouchoucha & Abd El Krim Souissi Abd El Kader Derdouche 

Robotics & Computer-integrated manufacturing Laboratory, Development and Research center, 

Military school Polytechnic(Ex: Enita), Blida, Algeria. 

BP 17c, Bordj El Bahri, Algiers, Algeria. 

E-mail: mouloud_bouchoucha@yahoo.fr 

Abstract 

A fuzzy sliding mode control (FSMC) schemes who has 

been proposed these last years is adopted as a solution 

for a single-input and multi-output benchmark system 

which is the inverted pendulum to handling instability, 

nonlinearity, uncertainly and coupling . The motivation is 

to combine the best features of the variable structure by 

sliding mode and fuzzy logic control to achieve rapid and 

accurate control. the number of the fuzzy rules is reduced 

by sliding surface and the chattering in the sliding mode 

is greatly reduced by the interpolation property of fuzzy 

control algorithm. Simulations and experiments to verify 

the schemes are presented. The results show a robust 

performance are attained by elimination of the chattering 

and rapid and accurate control. In the end the FSMC 

schemes are compared with the conventional sliding 

mode control design. 

Key words: Chattering, fuzzy logic control, fuzzy 

sliding mode control, identification, inverted pendulum, 

PID controller, variable structure control. 

1. Introduction 

The inverted pendulum is a SIMO unstable benchmark 

system characterized by nonlinearity, coupling, parameter 

fluctuations and external disturbances…etc. Variable 

structure control VSC(sliding mode control) is one of 

invariant control capable to handling of theses 

undesirable characteristics. This method is a nonlinear 

discontinuous control where the control law switch in a 

high frequency between two structures according to a 

switching(sliding) surface (combination of states system). 

Of the property of invariance with respect to the 

parametric variations and external disturbances, these 

system of control don’t require a high degree of 

accuracy in the identification of the parameters of the 

system to be controlled. However the chatter encountered 

in the VSC due to an oscillations of the control in high 

frequencies can excite unmodeled dynamics and renders 

this method impractical for most application. A 

continuous form of control approach (among others 

approaches) has been proposed replacing discontinuous 

form by a linear continuous in a boundary layer around 

the switching surface [Slotine 84]. Another solution has 

been appeared these last years : the fuzzy control by its 

interpolation characteristic combined with the sliding 

117 

mode control allows to reduce the control chattering. The 

results obtained during the application of this hybrid 

method of control to the inverted pendulum gave better 

results in simulation and in real time that those found by 

using the VSC in its discontinuous and continuous form. 

Although the order with variable structure has 

advantages, the phenomenon of chattering limits its 

implementation in real time. Indeed, due to the fast 

commutation of the order, the oscillations of high 

frequencies can excite dynamic not modelled system: 

This east can be harmful with the actuators. Several 

work was the subject of the development of techniques 

to reduce this undesirable phenomenon, among these 

last, one finds the method of the limiting band (Boundary 

Layer) of Slotine[11] where the discontinuous component 

is replaced in the vicinity of the hypersurface of slip by a 

continuous function. Another solution appeared these 

last years: the fuzzy order by its characteristic of 

interpolation combined with the order by mode of slip 

makes it possible to give better results to cure chattering. 

The results obtained during the application of this hybrid 

technique to the system of inverted pendulum gave better 

results in simulation and in real time that those found by 

using the VSC in its discontinuous form and the VSC in 

its continuous form. 

2. Variable structure control[1], [5] , [6], [7]. 

For a given system represented by : 

x& = A( 

x, 

t) 

+ B( 

x, 

t) 

⋅u 

(1) 

Where dim-x=n and dim-u=m 

How to solve the problem of the control of this system 

using the VSC? 

The resolution of this problem comes back to find : 

1) m switching functions represented in the vectorial 

form like S(x): 

S( 

x) 

n 

=∑c∑ i = 1 

n −1 

i⋅e i = en 

+ ci⋅e 

i ( cn 

i = 1 

= 1) 

With : ei = xi− 

xd 

: steady state error; 

xi : state of the system; 

xd: desired state; 

ci 

: parameters of the sliding surface. 

With a choice of the parameters of surface satisfying 

the desired performances, the stability problem is 

reduced to that of a linear system, therefore one can refer 

to the traditional criteria of stability. 

(2)

2) The variable structure control. 

⎧u+ 

( x, 

t) 

S( 

x) 

> 0 

U d( 

x, 

t) 

= ⎨ 

(3) 

⎩u−( 

x, 

t) 

S( 

x) 

< 0 

And such as the reaching or hitting mode satisfied the 

reaching condition (existence condition of the sliding 

mode) i.e. to reach S( 

x) 

= 0 in a finished time. 

Among the forms used, we find augmenting the 

equivalent control given by: 

U=U VSC = Ud +U eq = U eq - M.sign(S) (4) 

⎧ 1 S> 

0 

with : sign( 

S) 

= ⎨ 

⎩ −1 

S≤0 

Such as U eq is the equivalent control that allows to 

take the system on the sliding surface and it’s value is 

the average value of the discontinuous control during 

+ - 

commutation between its two values u and u . 

U eq is calculated posing S( 

x) 

= 0. 

In order to reduce the chattering phenomenon, the 

boundary layer solution method of Slotine was 

proposed[11]. In this method instead of a discontinuous 

form a continuous has been proposed and it’s given by : 

⎪⎧ 

− M. 

S S ≤Φ 

U = Ueq 

−M. 

sat( 

S) 

= Ueq 

+ ⎨ Φ 

( 5) 

⎪⎩ −M. 

sign( 

S) 

S > Φ 

3. Fuzzy sliding mode control(FSMC)[2], [3], 

[12]. 

The fuzzy sliding mode control is adopted as the 

second alternation for the reduction of the chattering 

phenomenon introduced by the discontinuous part (Ud) 

of the structure variable control. This hybrid control is a 

fuzzy control with the inputs of the controller are the 

sliding surface and its derivative : 

U = Ueq−U FSMC. 

sign( 

S) 

(6) 

With : U eq is the equivalent control. 

U FSMC is a Mamdani model(sets of fuzzy rules): with 

the sliding surface and its derivative as the input variables 

(sets of fuzzy rules): 

E J 

. 

RJ : If S is EJ and S is j Then UFSMC 

is AJ (7) 

E j 

. 

E . 

and A J are respectively fuzzy sets of the input 

. 

variables S and S and the output variable UFSMC 

of the 

fuzzy rule J. 

With the final control input U is calculated by the 

FSMC 

gravity center deffuzification method. 

4. Modeling of the system [8]. 

The system in question with the forces applied are 

represented on the figure 1. 

118 

θ 

x+l.sin(θ) 

Figure 1. Simplified model of the inverted pendulum. 

Such as: 

M : mass of the cart; 

m : mass of the pendulum >> mass of the stem; 

l : the length of the pendulum; 

F : the force applied to the cart by the DC servomotor; 

F : the force exerted by the cart on the pendulum; 

p 

θ : the angle which the pendulum with the vertical forms; 

x : position of the cart. 

b : coefficient of friction 

And as we don’t dispose a values of physical 

parameters of the system (parameters of the motor which 

actuates cart... etc), i.e. we cannot obtain a knowledge 

model of the system, it remains us to make simulations 

to identify the system. 

The system of the inverted pendulum is an unstable 

system in open loop, therefore the identification is done 

in closed loop after having to stabilize the system around 

its vertical axis by a simple PID controller determined 

experimentally. 

By raising of the input (U )and of output (θ, x) data and 

by the direct application of the algorithms of 

identification(Approach, ARMAX Structure, EP Method, 

MCS Criteria) [8] the model identified after validation is 

as follows: 

⎪⎧ 

x( n+ 

1) 

= Ax( 

n) 

+ Bu( 

n) 

⎨ 

(8) 

⎪⎩ 

y( 

n) 

= Cx( 

n) 

+ Du( 

n) 

Where ( n): 

x The state vector of dim ( x) 

= 4; 

( n): 

u The control vector of dim ( u) 

= 1; 

( n): 

m.g 

F 

y The output vector of dim ( y) 

= 2. 

With : 

⎡ 1.0018 0.0006559 

⎢ 0.010537 

A = ⎢ 

⎢ -0.01055 

⎢ 

⎣-0.013181 

1.0081 

-0.033191 

-0.10192 

b& x 

x x=0 

0.022802 -0.005164⎤ 

⎡-0.000328⎤ 

0.094702 -0.027377 

⎥ ⎢-0.003705⎥ 

⎥, 

= ⎢ ⎥ 

0.96423 -0.005293 

⎥ ⎢ 0.07789 ⎥ 

0.23494 0.91199 ⎥ 

⎦ 

⎢ 

⎣ -0.11482 

⎥ 

⎦ 

1 B 

M 

FP 

⎡ 2.8579 -0.00109 

0.038219 -0.016017 

⎤ ⎡0⎤ 

C = ⎢ 

⎥ , D= 

⎣0.16267 

-0.28894 

-0.029871 

-0.0054845 

⎢ ⎥ 

⎦ ⎣0⎦ 

This model will be used in simulation to locate initially 

the parameters of the controller before passing to the 

implementation in real time. 

l

5. Simulation Results 

5.1 Statement of the problem and definition of the 

sliding surface 

The problem such as it is posed relates to the control of 

a system at one input, the potential U () t applied to the 

DC servomotor ensuring the movement of the set cartpendulum 

and to two outputs : The angle θ () t between 

the pendulum and the vertical and the position of the cart 

x() 

t with regard to a fixed reference mark. In order to 

define the switching surface, we proposed in this work to 

use a variable y () t which includes the two variables x() 

t 

and, with a weighting on θ () t appreciably large, to give 

a priority of stabilization of the angle. The use of this 

variable was validated during the application of the 

traditional regulation of type PID. 

From where surface is defined as follows: 

S( 

y) 

= λy 

+ y& 

λ f 0 

= λ( 

x + aθ 

) + ( x& 

+ aθ& 

) (9) 

= λx 

+ x& 

+ λaθ 

+ aθ& 

With the state representation of our system : 

x & = Ax 

+ Bu 

(10) 

The state variables hosen are defined by posing the vector 

of state [ ] T 

= x x x x 

x 1 2 3 4 

x = x − x 

1 d 

x = x& 

− x& 

= −x& 

2 

d 

x3 

= θ d −θ 

= −θ 

x = θ& 

−θ& 

= −θ& 

4 

d 

And the surface S is the form : 

T 

T 

T 

S = K x ⇒ S& 

= K x& 

= K ( Ax 

+ Bu) 

(11) 

[ ] 

With: K a a 

(12) 

T 

= λ 1 λ 

The parameter a is selected equal to 3 in simulation 

and real time. 

5.2 Calculation of the Control input 

The control law of the system takes the form (4). 

With the equivalent control is calculated as follows: 

T 

S = 0 ⇒ S& 

= K ( Ax 

+ Bu) 

= 0 

(13) 

T −1 

T 

U = − K B K A 

eq 

( ) x 

And the discontinuous term takes the form: 

U d = −M. 

sign( 

S ) 

(14) 

Therefore the total control is written : 

T −1 T 

U = −( 

K B) 

K Ax 

− Msign( 

S ) (15) 

After several tests in simulation, and with the initial value 

of xd = 0m 

, the values chosen for λ and M in order to 

ensure the good performances are λ = 17 and M = 2. 

4 

(figure 2). 

The results obtained, with these parameters, show that 

the increase in λ improves the steady state error and the 

settling time that it is on the angle of the pendulum or the 

position of the cart. Whereas a large overshoot and a 

degradation of the settling time appear as of certain 

119 

values of λ . The phenomenon of chattering is visible on 

the response of the system. The oscillatory transitory 

mode present in the response of the cart justifies the 

priority of the stabilization of the angle of the pendulum 

compared to the control of the position of the cart. 

θ(rad) 

In order to reduce the phenomenon of chattering which 

is regarded as a real obstacle in the realization of the 

VSC, the continuous form of Slotine[11] is adopted (5). 

Simulation is made under the same conditions as those 

fixed in the preceding application that it is on the initial 

conditions or the parameters of surface and the control 

input so that we can make comparisons with Φ = 2. 

03 . 

θ(rad) 

0.25 

-0.25 

-0.5 

0 1 2 3 4 5 

4 

2 

Time (sec) 

The Sliding surface 

S 

0.5 

0.25 

-0.25 

0.5 

0 

0 

-2 

0 

The angle of the 

pendulum 

-0.5 

-0.3 

0 1 2 3 4 5 0 1 2 3 4 5 

Time (sec) 

Time (sec) 

S 

4 

2 

0 

-2 



pendulum 

x(m) 

U(V) 

-4 

0 1 2 3 4 5 

Time (sec) 

x(m) 

-4 

0 1 2 3 4 5 

Time (sec) 

-0.3 

0 1 2 3 4 5 

Time(sec) 

U (V) 

0 

-0.1 

-0.2 

10 The control input 

-10 

0 1 2 3 4 5 

Time (sec) 

Figure 3. Simulation results of the continuous form 

of the VSC(U = Ueq −M. 

sat( 

S ) ). 

According to the results obtained for the two forms of 

the control(discontinuous, continuous), one notices that 

the first form ensures a good speed for the system, in 

figure 2, the settling times of the cart and of pendulum 

are respectively: 1. 3s 

and 0 . 8s 

0.3 

0.2 

0.1 

0 

-0.1 

-0.2 

10 

5 

0 

-5 

0.3 

0.2 

0.1 

5 

0 

-5 

The position of the cart 

The Control 

-10 

0 1 2 3 4 5 

Time (sec) 

Figure 2. Simulation Results of the VSC 

(U = U eq−Msign 

S . 

( ) 

The position of the cart

On the other hand, from the energy point of view, the 

control input applied to the system is large even in 

steady state mode and present a considerable chattering 

which become null by using the continuous form of the 

VSC with the settling times of the cart and pendulum 

equal respectively to: 3s and 1. 

3s 

that confirms 

degradation in term of speed of the system (figure 3). 

The fuzzy sliding mode control FSMC (6) is adopted 

as a second solution for the chattering phenomenon. The 

control law becomes then nonlinear but continuous and it 

is given by a combination of fuzzy rules with the inputs 

is surface (9) and its derivative and the output is the 

control U FSMC as follows: 

If S is P and S is P Then U is NH; 

& FSMC 

If S is P and S is Z Then U is NB; 

& FSMC 

If S is P and S is N Then U is NM; 

& FSMC 

If S is Z and S is P Then U is NS; 

& FSMC 

If S is Z and S is Z Then U is Z; 

& FSMC 

If S is Z and S is N Then U is PS; 

& FSMC 

If S is N and S & is P Then U FSMC is PM; 

If S is N and S & is Z Then U FSMC is PB; 

If S is N and S & is N Then U FSMC is PH. 

With the memberships functions of S, S and U & 

FSMC 

chosen after several tests are given on the following 

figure: 

µ 

S 

N Z P 

-2 0 +2 

(a) 

S 

( c) 

µ U FSMC 

Simulation is made under the same conditions as those 

fixed in the preceding applications. 

The obtained results show(figure 5.) a fast convergence 

towards the stabilization position of the pendulum 0.9s 

and the cart 1.5s. Compared to the discontinuous form of 

VSC, this approach FSMC allows to reduce considerably 

the chattering phenomenon of (almost no one), with a 

small deterioration in term of performances(rapidity, 

overshoot) on the other hand this hybrid approach has a 

better speed than that obtained by VSC with a boundary 

layer. 

µ 

S & 

N Z P 

-40 0 +40 

(b) 

NH NB NM NS Z PS PM PB PH 

-1.5 -1.12 -0.75 -0.37 0 0.37 0.75 1.12 1.5 

S & 

U FSMC 

Figure 4. memberships Functions : (a), (b) sliding surface 

and its derivative, (c) FSMC 

120 

θ(rad) 

0.4 

0.2 


pendulum 

0.3 

0.2 

0.1 


0 

0 

-0.2 

-0.1 

-0.2 

-0.4 

0 1 2 3 4 5 

Time (sec) 

-0.3 

0 1 2 3 4 5 

Time (sec) 

S 

0 

-2 

x(m) 

4 


10 

2 

5 

-4 

0 1 2 3 4 5 

Time (sec) 

6. Experimental results 

-10 

0 1 2 3 4 5 

Time (sec) 

6.1 The interfacing system [ 4 ], [ 9 ] 

The system allows the interfacing and the treatment of 

the different inputs and the implementation of the control 

algorithms for the inverted pendulum is a dSPACE 

card(DS1102). 

From the programming point of view, the majority of 

the algorithms concerning the development of the control 

laws are realized by using MATLAB/SIMULINK 

software. This last allows fast implementation thanks to 

easy programming and to the use of specialized libraries: 

TOOLBOXS. 

6.2 Calculation of the control 

In all what follows, the conditions of the experimental 

tests are: 

♦ the sampling period is fixed at: T sap = 10ms 

♦ the initial conditions are fixed at: 

θ ( ) 0 =θ 0 = − 0. 

35 rad and x 0= x( 

0) 

= 0. 

21m, 

the 

choice of the initial conditions of different signs is 

due to the technical limitations of the system; 

♦ the best parameters of surface are given by the 

vector : 

T 

K = [ 12. 

07 1 36. 

21 3] 

For the first form of the control i.e. : 

U = Ueq 

−M. 

sign( 

S ) 

With : M = 1 it is the best value obtained after several 

tests and which ensures a certain speed to the system. 

The results obtained are represented on figure 6, with 

the initial conditions quoted previously and the final 

state θ () t = d 0 rad and x () t = d 0 m One notices a fast 

convergence of the position of the cart as well as the 

angle of the pendulum towards their references (null), 

the settling times of the system are approximately 0 . 7 s 

for the angle and 1s the cart (the angle of the pendulum 

U(V) 

0 

-5 

The control input 

Figure 5. Simulation results of the FSMC : 

U = Ueq−U 

FSMC. 

sign( 

S)

has priority that the position of the cart). In steady state 

mode discontinuous form commutates very quickly 

between its two limits, which influences the total control, 

where one notices a range of variation relatively large, 

the last generates fluctuations on the output level of the 

system as well as in the sliding surface, it’s the 

chattering phenomenon. 

( rad) 

θ 

Figure 6. Experimental results of the VSC 

(U = U − Msign ). 

eq 

( S ) 

In order to limit the chattering phenomenon the linear 

continuous form of Slotine is adopted(5). 

With: M = 1 and the value of the boundary layer 

Φ = 1. 

these values chosen after several tests ensure a 

certain speed for the system and decrease the chattering 

effect. 

θ(rad) 


pendulum 

-0.8 

Time (sec) 

Time (sec) 

-0.4 

0.8 

0.4 

0.45 

0.3 

0.15 


0 

0 

0 2 4 6 8 10 

-0.15 

-0.3 

-0.45 

0 2 4 6 8 10 

-8 


8 


0.8 

4 

4 

0.4 

S 

0.8 

0.4 

0 

-0.4 


pendulum 

x(m) 

-0.8 

0 2 4 6 8 10 

Time (sec) 

-8 


4 

S 

0 

-4 

-8 

0 2 4 6 8 10 

Time (sec) 

0 

-4 

x(m) 

-8 

0 2 4 6 8 10 

Time (sec) 

x(m) 

0.45 

0.3 The position of the cart 

0.15 

0 

-0.15 

-0.3 

-0.45 

0 2 4 6 8 10 

Time (sec) 

-8 


4 

Figure 7. Experimental results of the continuous form 

o f the VSC ( U = U eq−Msat(S)). 

The results obtained are represented on figure 7, for a 

final position θ () t = d 0 rad and x () t = d 0 m with the 

same initial conditions. One notices a fast convergence 

of the position of the cart as well as the angle of the 

pendulum towards their references, the settling times of 

the system are approximately 1. 2s 

for the angle and 1 . 3s 

for the cart. Compared with the results of the 

first(discontinuous) form, the system of inverted 

pendulum became slow but the chattering is reduced what 

is similar to the results obtained in simulation. 

U(V) 

U(V) 

0 

-4 

-8 

0 2 4 6 8 10 

Time (sec) 

0 

-4 

-8 

0 

2 4 6 8 10 

Time (sec) 

121 

The hybrid control FSMC is adopted as a second 

solution to the chattering phenomenon(6). 

The results obtained are represented on figure 8, for a 

final position θ d () t = 0 rad and xd () t = 0 m with the 

same initial conditions. One notices a fast convergence 

of the position of the cart as well as the angle of the 

pendulum towards their desired positions, the settling 

time is of 1s for the angle of the pendulum, and 1. 2s 

the 

cart. Compared with the results obtained with the 

discontinuous form of the VSC, the system converges 

less fast, with less oscillations and a very reduced 

chattering, whereas it is faster, with a small overshoot, 

that the VSC controller with a boundary layer. 

θ(rad) 

0 

-0.4 


pendulum 

-0.8 

0 1 2 3 4 5 6 

Time (sec) 

S 

8 

4 

0 

-4 

-8 0 


2 3 4 5 6 

Time (sec) 

x(m) 

Figure 8. Experimental results of the FSMC 

U = Ueq−U FSMC. 

sign( 

S) 

7. Conclusion 


-8 

0 1 2 3 4 5 6 

Time (sec) 

the variable structure control was presented and tested 

in simulation and in real time on a single-input and 

multi-output system type(benchmark) which is the 

inverted pendulum. The results obtained make it possible 

to conclude that the VSC in its discontinuous form could 

stabilize the system of the inverted pendulum and to 

make rapidly take its representative state up to the 

equilibrium point. However the chattering is significant 

as well as the furnished energy. Because of the 

chattering problem, the performances in term of precision 

are bad. In order to remedy of this phenomenon which is 

regarded as a real obstacle in the application of the 

VSC, the continuous form(boundary layer) of Slotine is 

adopted as a first solution and has allowed to reduce 

chattering but with a deterioration in term of settling 

time. The hybrid control with variable structure and by 

fuzzy logic which has been proposed these last years is 

adopted as a second solution, and with a good choice of 

the boundary layer(membership functions), this hybrid 

control was able to stabilize the system, to reduce the 

chattering and the overshoot and to give better 

performances compared to the method of the continuous 

form. 

Thus, in the application of the variable structure control 

it is necessary to make a compromise between the speed 

U(V) 

0.2 

0 

-0.2 

8 

4 

0 

-4 


0 1 2 3 4 5 6 

Time (sec)

of the system and the reduction of the chattering 

phenomenon and all depends on the applications. 

References 

[1] H. Buhler : Réglage par mode de glissement. Presses 

Polytechniques et Universitaires Romandes, 1986. 

[2] M. Bouchoucha : Conception d’un contrôleur à logique floue basé 

sur la théorie des modes glissants. thèse magister, EMP, 1999. 

[3] M. Bouchoucha, K. Belarbi : Full Design of Fuzzy Sliding Mode 

Controller : A Genetic Solution. IASTED (AIA 2001), Marbella, 

Espagne. 

[4] dSPACE : ControlDesk : Experiment Guide. dSPACE GmbH, 

Germany, 1999. 

[5] R.A. Decarlo, S.H. Zak et G.P. Matthews : Variable structure 

control of nonlinear multivariable systems : A tutorial . proc.IEEE, vol 

76, N° 3, pp-212 –232, 1988. 

[6] Emelyanov & Taran : Sur une classe de système de régulation 

automatique à structure variable. Thèse de doctorat de troisième cycle, 

Université de P.Sabatier, fevrier 1972. 

[7] J.Y. Hung, W.B. Gao, J.C. Hung : Variable structure control: A 

survey. trans.Ind.Electron, vol.40, N°1, pp 2-21, 1993. 

[8] P. BORNE, G. DAUPHIN-TANGUY, J.P. RICHARD, F. 

ROTELLA et I. ZAMBETTAKIS : Modélisation et identification des 

processus. Tome 2, Technip, 1992. 

[9] "Manuel d’utilisation", documentation de la maquette du système 

de pendule inversé. 

[10] H. S. Ramirez : Differential geometric methods in variable 

structure control. Int . J . Control , vol 48, N° 4 , pp. 1359-1390 , 1988. 

[11] J.J.E. Slotine : The robust control of robot manipulators. Int . J. 

Robotics, vol 4, 10.2, pp 49-64, 1985. 

[12] S. Kawaji, N. Matsunaga : Fuzzy control of VSS type and ist 

robustness. IFSA’91 Brusels, preprints vol, pp 81-88, 1991. 

122

ECAD/ECAE 2004, University of Durham, UK, November 15-16, 2004 

MODELING OF TRANSPORT FLOWS FOR INTERDISCIPLINARY SYSTEMS 

INTEGRATION 

Anatoly Levchenkov, Aleksandrs Suslenkovs 

Riga Technical University 

1, Kalku St., Riga, LV-1658, Latvia 

mailto:levas@latnet.lv 

Keywords: software agent, schedule , optimization of transport flows,. internet technologies 

ABSTRACT 

The work is based on the algorithm of optimization of transport flows in complicated managing systems. 

The aim of the work is a decomposing of the system and analysis of subsystems for the further 

optimization and correction of the information-transport flows using the methods of scheduling theory 

and decision making. 

At the same time it is supposed that there is no an interrelation between the decisions and an established 

order, i.e. the operations do not depend on their execution sequence. Most of the well-known methods of 

decision making with the help of design and analysis of correspondent operational models are 

investigated in the work. 

The transport flows in the work are presented as a system which needs an ordering of data with a 

reasonable usage during short time limits. The work suggests a modeling of schemes of flows routing 

with the further adaptation in a particular environment with the use of standards for interdisciplinary 

systems of integration. 

There is a developed on the obtained data mathematical model of a system of transport flows integration 

and representation in the form of totality of these flows factors, which could be demonstrated in a vector 

expression. The task of adaptation is an application of the elaborated algorithm of integration system in 

software agents and processing of information with the help of integrated data bases. The adaptation 

plays a role of deeper feedback (which has several hierarchies), improving the process of management of 

the complicated system namely transport flows. 

An additional solution of the problem of transport flows management is a development of special 

software which allows solving the problems in time. The work proposes to use Internet technologies and 

standards of system for joining of separate information systems for solution of specific problems caused 

by the fact that many processes cover several different information applications. 

123

ECAD/ECAE 2004, University of Durham, UK, November 15-16, 2004 

MODELING OF POWER DISTRIBUTION SYSTEMS WITH THE HELP OF 

INTERNET BASED DATA PROCESSING TEHNOLOGIES 

Anatoly Levchenkov, Diana Rihtere 

Riga Technical University 

1, Kalku St., Riga, LV-1658, Latvia 

mailto:levas@latnet.lv 

Keywords: software agent, schedule, electrical energy distribution,. internet technologies 

ABSTRACT 

During the work a software agent modeling has been observed for solving the problem of electrical 

energy distribution. The tasks of the work could be described in the following way: to model a power 

distribution system, which goes through all the points of electrical energy distribution, and which allows 

distributing it to all consumers with minimum expenses and time. To provide fast operation of the 

program and simplified location and filtration of information software agents have been applied. 

Scientific importance of the work is connected with application of the algorithms of incoming and 

processing of inquiries together with algorithms of graph theory. From the practical application point of 

view the results could be used in the management of the process of power distribution, providing the 

management of a particular unit in a day-night regime and decreasing the management expenses for manhour 

spent for decision making in states of emergency in order to apply the possibilities of systems of the 

Internet data bases as well as to model operation of the real power distribution systems. 

Software agents are used for providing fast operation of the program, location of the information and 

simplifying of the filtration. To display the data on consumers a data base has been elaborated in the 

Internet environment, which represents the information on payments and time resources of each 

consumer that is necessary for delivery of electrical energy. The elaborated algorithm is described and a 

logistic practical example of application of the algorithm is considered. 

There is a wide area of possibilities to solve the problems of power distribution. The optimization of 

power distribution can provide also an improvement in operation of another system. In present economic 

conditions a solving of power distribution problem, based on the optimization of electric transport 

operation, is vary actual. One of the solutions is application of algorithms of theory of scheduling. It 

seems reasonable to supply of each type of electric transport with electric energy based on real needs. 

There are real possibilities to use the mentioned above algorithms of graph theory, applied for an optimal 

route defining, with the algorithms of theory of scheduling, which allow optimizing the schedule of the 

vehicle movement. The newly-elaborated combination of the methods will help to decrease consumption 

of electric energy, expenses of electric transport running, coordinating in its turn the routes and schedules 

of the transport. The data could be input into the Internet with the help of software agents. Applying the 

suggested Internet technologies for data processing it is possible to find the optimal solution of the 

problem. 

124

Author Index 

A 

Almaini, A 95 

Arumugam, J. 12 

B 

Baguley, P. 41, 101, 106 

Bouchoucha, M. 117 

C 

Cooke, M. D. 51 

D 

Derdouche, A. 117 

Desoi, J.C. 36 

G 

Gallant, A. J. 67 

Graetz, F. 2 

Grebenstein, K. 111 

H 

Hackney P.D 75 

I 

Ito, T 6 

125 

L 

Lenders, M. 36 

Levchenkov A. 123,124 

M 

Mankel, A. 2 

Maropoulos, P.G. 59 

Mukherjee, D. 80 

N 

Neerukonda, N. 12 

P 

Page, T. 101,106 

Pleßow, M. 90 

R 

Rihtere, D. 124 

Rolt, S. 59 

S 

Schaefer, D. 41, 46, 101,106 

Schuh, G. 36 

Schumann, C.A. 111 

Solano, B. 59 

Šormaz, D. 12 

Souissi, A. 117 

Stube, B. 90 

Suslenkovs, A. 123 

T 

Tittmann, C. 111

V 

Vigerske, W. 90 

W 

Ward, T. 21 

Witte, V. 36 

Wood, D. 51, 59, 67 

126 

Y 

Yang, M. 95 

Z 

Zaeh, M. F. 2 

Zughaid, N.S. 75

Searchable Index 

The electronic version of these proceedings allows you to perform full text searches of the 

included papers. 

Search 

127

ECAD/ECAE2004 - the Systems Realization Laboratory

Create successful ePaper yourself

Delete template?

Save as template?