jaes - vol.10 - no.3

Journal of Applied Engineering Science 10(2012)2

I M P R E S S U M

J O U R N A L O F A P P L I E D E N G I N E E R I N G S C I E N C E (J A E S)

The journal publishes original and review articles covering the concept of technical science, energy and environ-ment, industrial engineering, quality management and other related sciences. JAES is Open-Access Journal that follows new trends and progress proven practice in listed fi elds, thus creating a unique forum for interdisciplinary or multidisciplinary dialogue. JAES is part of the electronic journal editing with a transparent editorial and review policy. Provided are:

Online paper submission and tracking of review process with communication between editors, authors and reviewersCrossRef: assignment of numerical identifi ers (DOI) to assure greater visibility and accessibility of journal articlesCrossCheck: control for originality of submitted papers, to prevent plagiarism and duplicate publicationsKWASS: automatic extraction of keywords from disciplinary thesaurusOnline-fi rst publishingAutomatic transfer of metadata to SCIndeks that support international protocols for data transfer

All published articles are indexed by international abstract base Elsevier Bibliographic Databases through service SCOPUS since 2006 and through service SCImago Journal Rank since 2011.Serbian Ministry of Science admitted the Journal of Applied Engineering Science in a list of reference journals. Same Ministry fi nancially supports journals publication.

PublisherInstitute for Research and Design in Commerce and Industry - IIPP; www.iipp.rsFor publisher: Prof. dr Branko VasiCopublisherFaculty of Transport and Traffi c Engineering Belgrade University; www.sf.bg.ac.rsFor copublisher: Prof. dr Slobodan GvozdenoviEditor in ChiefProf. dr Jovan TodoroviFaculty of Mechanical Engineering, Belgrade;Assistant EditorDr Predrag Uskokovi, IIPPEditorial Board Prof. dr Gradimir Danon, Faculty of Forestry, Belgrade; Doc. dr Duan Milutinovi, Institute for Transport and Traffi c CIP, Belgrade; Mr ore Milosavljevi, CPI - Process Engineering Center, Belgrade; Prof. dr Miodrag Zec, Faculty of Philosophy, Belgrade; Prof. dr Nenad aji,Mining and Geology Faculty, Belgrade; Prof. dr Vlastimir Dedovi, Faculty of Transport and Traffi c Engeneering, Belgrade;Prof. dr Mirko Vujoevi,Faculty of organizational sciences, Belgrade;Doc. dr Vladimir Popovi,Faculty of Mechanical Engineering, Belgrade;Doc. dr Vesna Spasojevi Brki,Faculty of Mechanical Engineering, Belgrade.ISSN 1451-4117 UDC 33Papers are indexed by SCOPUSJournal of Applied Engineeering Science is available at: www.engineeringscience.rs http://scindeks-eur.ceon.rs/index.php/jaeshttp://www.singipedia.com/content/1166-naucni-casopisiDesigned and prepress: IIPP

International Editorial BoardProf. dr Vukan Vui,University of Pennsylvania, USA;Prof. dr Robert Bjekovi, Hochschule Ravensburg-Weingarten, Germany;Prof. dr Jozef Aronov, Research Institute for Certifi cation JSC, Russia;Prof. dr Jezdimir Kneevi, MIRCE Akademy, England;Dr Neboja Kovaevi, Geotechnical consulting group, England;Adam Zielinski, Solaris Bus & Coach, Poland;Prof. dr Milo Kneevi, Faculty for Civil Engineering, Montenegro;MSc Sinia Vidovi, Energy Testing & Balance Inc, USA;Dr Zdravko Milovanovi,Faculty of Mechanical Engineering, Banja Luka.Publishing CouncilProf. dr Milorad Milovanevi,Faculty of Mechanical Engineering, Belgrade; Milutin Ignjatovi,Institute for Transport and Traffi c CIP, Belgrade; Dragan Beli,Transport Company Lasta, Belgrade;Dr Deda elovi, Port of Bar, Bar;Dr Drago erovi, Adriatic Shipyard, Bijela;Cvijo Babi,Belgrade Waterworks and Sewerage, Belgrade;Nenad Jankov, Power Plant Kostolac B, Kostolac;Miroslav Vukovi, Mercator Business System, Belgrade;Duan uraevi, Euro Sumar, Belgrade.Editorial Offi ceNada Stanojevi, Milo Vasi, Darko Stanojevi,Milo Dimitrijevi, Mirjana Solunac, Ivana Spasojevi, IIPP, Belgrade;Bojan Mani, Faculty of Mechanical Engineering, Belgrade.Printed by: Sigrastar, Beograd

Institute for research and design in commerce & industry, Belgrade. All rights reserved. Journal of Applied Engineering Science 10(2012)3

C O N T E N T S

Dr Goran PutnikADVANCED MANUFACTURING SYSTEMS AND ENTERPRISES: CLOUD AND

UBIQUITOUS MANUFACTURING AND AN ARCHITECTURE127 - 134

MSc Bojan Jovanovski, Dr Robert Minovski, Dr Siegfried Voessner, Dr Gerald Lichtenegger

COMBINING SYSTEM DYNAMICS AND DESCRETE EVENT SIMULATION-OVERVIEW OF HYBRID SIMULATION MODELS

135 - 142

Dr Isabel L. NunesFUZZY SYSTEMS TO SUPPORT INDUSTRIAL ENGINEERING MANAGEMENT 143 - 146

Dr Mirjana Misita, Dr Galal Senussia, MSc Marija MilovanoviA COMBINING GENETIC LEARNING ALGORITHM AND RISK MATRIX MODEL

USING IN OPTIMAL PRODUCTION PROGRAM147 - 152

Jelena Jovanovi, Dr Dragan Milanovi, Dr Mili Radovi, Radisav ukiINVESTIGATIONS OF TIME AND ECONOMIC DIMENSIONS OF THE COMPLEX

PRODUCT PRODUCTION CYCLE153 - 160

Dr Vidosav MajstoroviTOWARDS A DIGITAL FACTORY - RESEARCH IN THE WORLD

AND OUR COUNTRY161 - 165

Dr Jezdimir KneeviTIME TO CHOOSE BETWEEN SCIENTIFIC AND ADMINISTRATIVE

APPROACH TO RELIABILITY 167 - 173

EVENTS REVIEW 174

ANNOUNCEMENT OF EVENTS 175 - 176

BOOK RECOMMENDATION 177

INSTRUCTIONS FOR AUTHORS 178 - 179

EDITORIAL AND ABSTRACTS IN SERBIAN LANGUAGE 180 - 184


E D I T O R I A L

PREFACE TO SPECIAL TOPIC: SELECTED PAPERS FROM THE 5TH INTERNATIONAL SYMPOSIUM OF INDUSTRIAL ENGINEERING SIE 2012

President of the SIE 2012 Organizing Committee

The 5th International Symposium of Industrial Engineering SIE 2012, held in June, 2012 in Belgrade, Serbia, was aimed at providing a unique platform to meet frontier researchers, scientists, as well as practitioners and share cutting-edge developments in the fi eld. The SIE 2012, fi fth in the series of SIE meetings, was organized by Industrial Engineering Department, Faculty f Mechanical Engineer-ing, University of Belgrade, Serbia and Steinbeis Advanced Risk Technologies Stuttgart, Germany.

The Symposium also fostered networking, collaboration and joint effort among the conference partic-ipants to advance the theory and practice as well as to identify major trends in Industrial Engineering today. Proceedings with over 70 papers and disscusions by more then 160 authors have contributed to better comprehension the role and importance of Industrial Engineering in this country, both in do-main of scientifi c work and everyday practice. Despite the widely varying backgrounds and interests of the participants, the schedule of the meeting kept them all fully engaged by providing a platform where ideas at the cutting edge of industrial engineering could be exchanged with great enthusiasm. We think that the synergies and international collaborations resulted through sharing of knowledge and cross-fertilization of ideas at the earlier SIE meetings led to greater maturity at the SIE 2012.

This traditional symposium has made a great contribution to improving awareness of the importance of industrial engineering at the international, national and local levels. The presentations, which were organized during the symposium, presented the achievement at all levels of scientifi c research through a systematic approach, to specifi c practices, both in large systems and small and medium enterprises. Also, further development of cooperation in the fi eld of industrial engineering with the aim of realization of a larger number of projects of international domain is necessary. Great dynamics trends in the fi eld of industrial engineering require expertise and wisdom to preserve the long-lasting knowledge base of industrial engineering, together with establishment of fl exibility and customiza-tion, that new challenges brought by the times in which we live and do business.

Issues number 3 and 4 of the Journal of Applied Engineering Science contain a selection of papers on all aspects of interdisciplinary themes treated in the SIE 2012. There are 11 papers that cover top-ics in the fi elds such as quality management, production management, risk management, project`s appraisal etc. The most relevant contributions are selected after a standard review process by disciplinary experts.

We give thanks to the members of SIE Scientifi c Committee and authors, as well to our sponsors. Special thanks are due to the staff of the Journal of Applied Engineering Science and all the referees for their careful work.Belgrade, June 2012

Prof. dr Vesna Spasojevi-Brki

Paper number: 10(2012)3, 229, 127-134

ADVANCED MANUFACTURING SYSTEMS AND ENTERPRISES: CLOUD AND UBIQUITOUS

MANUFACTURING AND AN ARCHITECTUREDr Goran Putnik *University of Minho, Faculty of Engineering, Braga, Portugal

doi:10.5937/jaes10-2511

In this paper, in the fi rst part an introduction to development of the concepts of Ubiquitous and Cloud Manufacturing is presented, as a model of advanced manufacturing systems and enterprises. In the second part an architecture, that might guide the implementation and exploitation of the Ubiquitous and Cloud Manufacturing is presented through an informal and conceptual presentation.Key words: Ubiquitous, Manufacturing systems, Enterprises, Clouds, Architecture, Paradigm

INTRODUCTION

The traditional Manufacturing was supersed-ed. The new dynamic and global business model forced traditional production processes to change in the sense of to be integrated in a global chain of resources and stakeholders. The agility and quick reaction to market changes is essential, and the high availability and capacity to effectively answer to requirements is one of the main sustainability criterion.

Globalization, innovation and ICT are trans-forming many sectors to anywhere, anytime plat-forms, towards an intelligent business model under design anywhere, make anywhere, sell anywhere paradigm [03]. We would add any-time too. Traditional suppliers and customers are transformed in services, where supplying or using profi les are a question of needs or con-text. One service (a Calculator, for instance) can execute (supply) something using other services (Add, Sub, Mult and Div operations) [15].

All these performances are considered on Ubiq-uitous and Cloud Manufacturing [08, 09, 18] sug-gest a manufacturing versionof ubiquitous and cloud computing (respectively) ubiquitous and cloud manufacturing and manufacturing with direct adoption of ubiquitous and cloud comput-ing technologies. In this context, resources are seen as services, essentially. This manufacturing service-oriented network can stimulate production-oriented to service-oriented manufacturing [01].

Many of existent infra-structures are already ubiquitous and/or cloud based or are changing

towards these virtual architecture. To use ef-fi ciently those infra-structures the applications must be transformed and follow services orient-ed applications pattern.

In this paper, in the fi rst part an introduction to development of the concepts of Ubiquitous and Cloud Manufacturing is presented, as a model of advanced manufacturing systems and enter-prises. In the second part an architecture, that might guide the implementation and exploitation of the Ubiquitous and Cloud Manufacturing is presented through an informal and conceptual presentation.

MANUFACTURING AS SERVICE SYSTEMS

Industrial and Product-Service Systems (IPS2) represents a paradigm shift from the separated consideration of products and services to a new product understanding consisting of integrated products and services creates innovation poten-tial to increase the sustainable competitiveness of mechanical engineering and plant design. The latter allows business models which do not fo-cus on the machine sales but on the use for the customer e.g. in form of continuously available machines. The business model determines the complexity of delivery processes. Characteristics of Industrial Product-Service Systems allow cov-ering all market demands [05]. Figure 1 shows service offer of Mori Seiki, while Figure 2 and Figure 3 shows types of Product-Service Sys-tems and scientifi c fi elds of action respectively.

* Faculty of Engineering, Largo do Capo, 4704-553, Braga; [email protected] Paper presented at the SIE 2012

127


Dr Goran Putnik - Advanced manufacturing systems and enterprises: cloud and ubiquitous manufacturing and an architecture

Figure 1: Service offer of Mori Seiki [06, 05].

Figure 2: Types of Product-Service Systems [06,05]

Figure 3: Scientifi c fi elds of action [05]

UBIQUITOUS SYSTEMS

Ubiquity is a synonym for omnipresence, the property of being present everywhere (Wikipedia). The state or quality of being, or appearing to be, everywhere at once; actual or perceived omnipresence: the ability to be at all places at the same time; usually only at-tributed to God (Wiktionary).

According to Weiser (1993) Ubiquitous Comput-ing represents: Long-term the PC and worksta-tion will wither because computing access will be everywhere: in the walls, on wrists, and in scrap computers (like scrap paper) lying about to be grabbed as needed.

128 , 229



Theredore, Ubiquitous Manufacturing Systems and Enterprises concept is related to the avail-ability of management, control and operation functions of manufacturing systems and enter-prises anywhere, anytime, using direct control, notebooks or handheld devices. It is related with Ubiquitous Computing Systems.Ubiquitous Manufacturing Systems (UMS), therefore, implies ubiquity of three general types of resources in organizations:

material processing resources (e.g. machine tools and other manufacturing/production equipment as resources), information processing resources (e.g. com-putational resources includes hardware and software), and knowledge resources (i.e. human resources, considering the humans as unique resources for knowledge generation and new products and services creation, and, at the end, the ultimate effectiveness of organizations).

However, there are two quite different approach-es to the concept of UMS.

The fi rst concept, considers ubiquity of the MS based on, i.e. uses, the ubiquitous com-putational systems (UCS), Figure 4.a, The second one which is original our ap-proach, considers ubiquity of the MS as a homomorphism , i.e. it is a mapping, of the ubiquitous computational systems (UCS), Figure 4.b [08, 09, 10].

Figure 4: a) UMS has UCS as an operating system only Ubiquity of Computational resources only; b) UMS operates as UCS Ubiquity of all Resources: Material processing, Knowledge, and Computational

resources [10]

Computing technology has evolved up to the point when Ubiquitous Computing System devel-opment and operation are possible, using pres-ent network devices, protocols and applications.From the other hand, ubiquity has been ad-dressed in relation to manufacturing systems as well. In (Foust, 1975) [04] the term ubiquitous is explicitly defi ned to be functional in an empiri-cal context The types of manufacturing which are both market oriented and have a frequency of occurrence greater than a specifi c limit which can be empirically defi ned are ubiquitous. .Foust (1975) cites Alfred Webers defi nition of ubiquitous manufacturing too: Ubiquity naturally does not mean that a commodity is present or producible at every mathematical point of the country or region. It means that the commodity is so extensively available within the region that, wherever a place of consumption is located, there are opportunities for producing it in the vicinity. Ubiquity is therefore not a mathematical, but a practical and approximate, term (praktisch-erNaherungsbegriff).

To the above defi nitions (by (Foust, 1975) and (We-ber, 1928)), [16] which consider ubiquity of resourc-es anywhere, we add the ubiquity in time any-time, which (the anytime), from its side, implies the dynamic, on-line, seamless, enterprises organi-zational and manufacturing system networking and reconfi gurability, or adaptability, that requires new organisational architectures and meta-enterprise organizations as creating and operating environ-ments, makes the UMS a true new paradigm.

The similar idea was referred in (Murakami &Fu-jinuma; 2000), (ref. in (Serrano & Fischer; 2007)). This approach is referred as well as Ubiquitous

networking that emphasises the possibility of building networks of persons and objects for sending and receiving information of all kinds

129, 229


and thus providing the users with services any-time and at any place.

The hypothesis is that UMS should be based on a hyper-sized manufacturing network, consist-ing of thousands, hundreds of thousands, or mil-lions of nodes, i.e. of manufacturing resources units, freely accessible and independent, Figure 5.

Further implications are that UMS manufacturing units should be, in the limit,primitive, i.e. individuals, or individual companies, and individually owned head-wear/software resources, Management and operation of UMS should ne informed by the discipline of chaos and complexity management in organizations, e.g. Chaordic System Thinking (CST) model [02]

1)

2)

Specifi c instruments should be used, such as meta-organizations (e.g. Market of Re-sources model), brokering and virtuality,These UMS hyper-sized manufacturing networks could be seen as manufacturing resources Internet of Things,These UMS hyper-sized manufacturing networks could be seen as manufacturing production social networks,These UMS hyper-sized manufacturing networks form and use clouds.

CLOUD BASED PLATFORM

Presentation of the cloud is transcribed from (Schubert L., ) - as the reference source cre-ated within the EC initiative and therefore it is the most relevant for an advanced Manufactur-ing Systems and/or enterprise.

3)

4)

5)

6)

Figure 5: Figurative presentation of VE evolution: from conservative, minimal network domain (a), towards ubiquitous network domain (d)

A cloud is a platform or infrastructure that en-ables execution of code (services, applications etc.), in a managed and elastic fashion, whereas managed means that reliability according to pre-defi ned quality parameters is automatically ensured and elastic implies that the resources are put to use according to actual current require-ments observing overarching requirement defi ni-tions implicitly, elasticity includes both up- and downward scalability of resources and data, but also load-balancing of data throughput.

Cloud has a number of particular characteris-tics that distinguish it from classical resource and service provisioning environments: (1) it is (more-or-less) infi nitely scalable; (2) it provides one or more of an infrastructure for platforms, a platform for applications or applications (via ser-vices) themselves; (3) thus clouds can be used for every purpose from disaster recovery/busi-ness continuity through to a fully outsourced ICT service for an organisation; (4) clouds shift the costs for a business opportunity from CAPEX to OPEX which allows fi ner control of expenditure

130


, 229


and avoids costly asset acquisition and mainte-nance reducing the entry threshold barrier; (5) currently the major cloud providers had already invested in large scale infrastructure and now offer a cloud service to exploit it; (6) as a con-sequence the cloud offerings are heterogeneous and without agreed interfaces; (7) cloud provid-ers essentially provide datacentres for outsourc-ing; (8) there are concerns over security if a busi-ness places its valuable knowledge, information and data on an external service; (9) there are concerns over availability and business continu-ity with some recent examples of failures; (10) there are concerns over data shipping over an-ticipated broadband speeds.Concerning the EU policy towards clouds, the document refers two main recommendations:Recommendation 1: The EC should stimulate research and technological development in the area of Cloud ComputingRecommendation 2: The EC together with Member States should set up the right regula-tory framework to facilitate the uptake of Cloud computingConcerning the types of clouds, for an advanced Manufacturing Systems and/or enterprise, the most important are the concepts of cloud types: (1) IaaS - Infrastructure as a Service, (2) PaaS - Platform as a Service, (3) SaaS - Software as a Service, and collectively *aaS (Everything as a Service) all of which imply a service-oriented architecture.

AN OVERALL SYSTEM ARCHITECTURE FOR ADVANCED MANUFACTURING

Advanced manufacturing system architecture, Figure 6, is a cloud based architecture that rep-resents the manufacturing system as a service system, integrating the services for

Real-time Data Acquisition Services for real-time data acquisition from the equipment through the embedded intelligent information devices services type/group Equipment In-telligent Monitoring Systems,Product Design Services, that integrates four environments: 1) Computer Aided Design, 2) Product data repository with embedded In-telligent System for Decision Making (for ac-cessing all relevant data, actual and historic as well as data analysis) from the equipment in use, 3) Mixed-reality Environment, and 4) Co-Creation (Collaborative) Environment for co-creative design services type/group Product Design Services;Equipment Operation Services, that inte-grates four environments: 1) Equipment Data Real-time with embedded Intelligent System for Decision Making, that provides all relevant data, actual and historic as well as data analysis and management suggestions, necessary for the productuion management

1)

2)

3)

Figure 6: Overall System Architecture for development, implementation and validation

131


, 229


2) Management environment, for monitor-ing, scheduling and controlling management activities, with embedded Intelligent System for Decision Making, 3) Mixed-reality Envi-ronment, and 4) Co-Creation (Collaborative) Environment for co-creative management services;The cloud infrastructure, that will provide the 1) infrastructure for the manufacturing system applications of all three types of resources: material processing resources, information processing resources (i.e. computational re-sources), and knowledge resources in the form of IaaS - Infrastructure as a Service; 2) platform for the manufacturing system appli-cations in the form of PaaS - Platform as a Service, and 3) manufacturing system soft-ware business applications in the form of SaaS - Software as a Service.

ICT Platform ArchitectureThe logical architecture of the ICT Platform is architecturefor integration of Representation, Mixed-reality representation, Real-time man-agement model, and Communication for col-laborative management.

4)

It is basically a 3.tier layer architecture consist-ing of (1) Presentation Layer, (2) Business Layer and (3) Data Layer.The Presentation Layer represents/defi nes ap-plications and support for all interfaces, views, presentations and communications for users.

The Business Layer represents/defi nes applica-tions and support for all business applications such as Decision Making applications, Intelligent System applications, Services Workfl ows.

The Data Layer represents/defi nes applications and support for all applications for data reposito-ry and management, including knowledge bases (e.g. for Intelligent System on the upper level).For each layer the corresponding technology to be employed is referred.

Co-Creation and Semiotics and Pragmatics platformAdvanced manufacturing system architecture will integrate environments, or so-called, co-creative platforms, for three co-creative environments: 1) for product design processes, 2) for operation, or production, management processes, and 3) for integrated design-production processes.

Figure 7: Advanced manufacturing system co-creative platform, for three co-creative environments: 1) for product design processes, 2) for operation, or production, management processes, and 3) for integrated

design-production processes

132


, 229


It means that the co-creative processes both group of agents will perform independently, i.e. the designers will be capable to perform their processes in their own environment separately from the managers 1st Co-Creative cycle, and the managers will be be capable to perform their processes in their own environment separately from the designers 2nd Co-Creative cycle,. However, additionally, both groups will be ca-pable to perform their processes jointly in a fully integrated and systemic way 3rd Co-Creative cycle, Figure 7.

The supporting technique will be the multi-user video-conferencing with auxiliary functionalities. A vision is presented on the Figure 8. These three cycles, and the video-conferencing environment, will provide full semiotic/pragmat-ics effects and support in order to enhance to maximum the cognitive and creative capacities of the participants, and a full co-creative, or co-design or co-evolving, and truly systemic envi-ronment.

SustainabilityThe three aspects of sustainability: economic, environmental and social should be implement-ed in the following way: Figure 7 and Figure 8.

Economic and environmental sustainability: Economic and environmental sustainability will be based on implementation of specifi c software-modules, with corresponded analytical models, for continuous evaluation of energy consumption and costs, environmental pollution and associated costs.

These models and applications will be embed-ded in data acquisition services, see the System Architecture, Figure 15.

Social sustainability: Advanced manufacturing system components will support Social sustain-ability goals enabling The creation of new jobs This effect will be possible because the advanced manufacturing system is conceived as a service system meaning a great degree of openness for performing these services, the maintenance management and design services, by individu-als (free-lancers), micro and small companies, that would form a dynamic network of services providers. In this way a potential for new jobs creation will be dramatically increased.

CONCLUSIONS

The architecture presented is of a general na-ture andopen in various aspects, with structural elements, in nature and in number, that enables development of an advanced manufacturing sys-tem or enterprise on different complexity levels which is on of the primary requirements for the capacity of achieving sustainability. Therefore, the architecture presented may have a number of implementation forms.It would be useful to remind that a number of under-lying technologies should be considered, and which were not possible to analyze due to the papers limited space. E.g. embedded intelligent informa-tion devices, real-time management (and design), mixed reality and augmented reality, semiotics and pragmatics, co-creation, chaos and complexity management, the theory of sustainability, web 2.0 to web 4.0, and others. In short, many of technologies are already present. However, from the other hand, there is a number of open technical, organizational and conceptual problems that requires hard work in the future. Two of the virtually most important prob-lems to work on are the interoperability, or integra-tion, of the Ubiquitous and Cloud Manufacturing and their adoption in society (and industry of course).

Figure 8: A vision of the multi-user video-conferencing system as the co-creative environment

133


, 229


ACKNOWLEGMENTS

The authors wish to acknowledge the support of: 1) The Foundation for Science and Technology FCT, Project PTDC/EME-GIN/102143/2008, Ubiquitous oriented embedded systems for globally distributed factories of manufacturing enterprises, 2) EUREKA, Project E! 4177-Pro-Factory UES

REFERENCES

Cheng, Y., Tao, F., Zhang, L., Zhang, X., Xi, G. H., & Zhao, D. (2010).Study on the utility model and utility equilibrium of resource ser-vice transaction in cloud manufacturing. Pa-per presented at the Industrial Engineering and Engineering Management (IEEM), 2010 Eijnatten F., Putnik G., Sluga A. (2007) Chaordic Systems Thinking for Novelty in Contemporary Manufacturing, CIRP Annals, Vol 56, No 1, pp. 447-450Elliott, L. (2010). The Business of ICT in Manufacturing in Africa: AfribizFoust, Brady J. (1975) Ubiquitous Manufac-turing, Annals of the Association of American Geog-raphers, Vol. 65, No. 1 (March 1975), pp. 13-17Meier H., Roy R., Seliger G. (2010) Industrial Product-Service SystemsIPS2, CIRP Annals - Manufacturing Technology, 59 (2010) 607627Mori Seiki CO., LTD, Service/Support von A-Z mit der Sicherheit des Herstellers. Service Brochure published by Mori SeikiMurakami, T., Fujinuma, A. (2000).Ubiqui-tous networking: Towards a new paradigm. Nomura Research Institute Papers, No. 2Putnik G. et al. (2004) Cells for Ubiquitous Pro-duction Systems, Proposal for R&D Project, Project reference: POSC/EIA/60210/2004, submitted to Fundaopara a Cincia e a Tecnologia (FCT), Lisbon, PortugalPutnik G. et al. (2006) Ubiquitous Production Systems and Enterprises - advanced enter-prise networks for competitive global manu-facturing, Proposal for R&D Project, Project reference: PTDC/EME-GIN/72035/2006, submitted to Fundaopara a Cincia e a Tecnologia (FCT), Lisbon, PortugalPutnik G.D., Cardeira C., Leito P., Restivo F., Santos J., Sluga A., Butala P. (2007) To-wards Ubiquitous Production Systems and Enterprises, in Proceedings of IEEE Int. Symp.on Ind. Electronics - ISIE 2007, Vigo, Spain

1)

2)

3)

4)

5)

6)

7)

8)

9)

10)

Putnik, G. D. (2010). Ubiquitous Manufactur-ing Systems vs. Ubiquitous Manufacturing Systems: Two Paradigms. In Proceedings of Proceedings of the CIRP ICME 10 - 7th CIRP International Conference on Intelligent Computation in Manufacturing Engineering - Innovative and Cognitive Production Tech-nology and SystemsPutnik, G. D., & Putnik, Z. (2010). A semiotic framework for manufacturing systems inte-gration -Part I: Generative integration model.International Journal of Computer Integrated Manufacturing, 23: 8, 691 - 709Schubert L. (2010) The future of cloud com-puting opportunities for European cloud com-puting beyond 2010, European Commission Information Society and MediaSerrano V., Fischer T. (2007) Collaborative innovation in ubiquitous systems, J Intell-Manuf (2007) 18:599615Usmani, S., Azeem, N., &Samreen, A. (2011). Dynamic Service Composition in SOA and QoS Related Issues International Journal of Comput-er Technology and Applications, 2, 1315-1321Weber A. (1928), Theory of the Location of Industries, translated by C. J. Friedrich (Chi-cago: University of Chicago Press, 1928), p. 51 (emphases by Foust, Brady J. (1975))Weiser,http://www.ubiq.com/hypertext/weiser/UbiHome.html, Xerox PARC Sandbox Server.Xu, X. (2012). From cloud computing to cloud manufacturing. Robotics and Computer-Inte-grated Manufacturing(28), 75-86

11)

12)

13)

14)

15)

16)

17)

18)

Paper sent to revision: 28.08.2012.Paper ready for publication: 27.09.2012.

134


, 229

Paper number: 10(2012)3, 230, 135-142

COMBINING SYSTEM DYNAMICS AND DISCRETE EVENT SIMULATIONS - OVERVIEW OF HYBRID

SIMULATION MODELSMSc Bojan Jovanovski *Ss. Cyril and Methodi University, Faculty of Mecahical Engineering, Skopje, MacedoniaDr Robert MinovskiSs. Cyril and Methodi University, Faculty of Mecahical Engineering, Skopje, MacedoniaDr Siegfried VoessnerInstitute of Engineering and Business Informatics, TU Graz, AustriaDr Gerald LichteneggerInstitute of Engineering and Business Informatics, TU Graz, Austria

doi:10.5937/jaes10-2512

Simulation and modelling has been widely accepted as one of the most important aspects of the Industrial engineering. The application and use of simulation models has grown exponentially since the 1950 until today. Over the years, the complexity of the simulated aspects has been adapted to the complexity of the analysed cases which has risen proportionally too. That is why techniques used many years ago, can often not give an adequate representation of the real world any more. For that reason, we propose to use hybrid simulation models, which are a combination of simula-tion paradigms in order to cope with this problem. In this paper, we will give an overview of selected researches and applications with an emphasis on Discrete Event Simulation and System Dynamics, as one of the core simulation based techniques in that area.Key words: Hybrids, Models, Simulation of models, System dynamics, Discrete-event simulation

INTRODUCTION

The advances in Industrial Engineering (IE) have gone a long way since the early beginnings and the experiments of Taylor, Gilbreth, Babbage, Towne and others. Not so much in the area of the fi eld, but in the direction of tackling even the smallest details possible. In order to do this the complexity of the problems grew, with that the data needed to be obtained and processed was also getting bigger. The computers played huge factor in keeping the Industrial Engineering alive and constantly being in trend. Not only because of the hardware possibilities and the calculations that could have been made now, but also from the point of view that many software packages have been developed in order to solve some kind of an IE problem. There are solutions for fi nding an optimal layout, managing production processes, tackling ergonomic issues, calculat-ing cost/profi t etc. (the intention is not to name vendors here).

Simulation and modelling has been widely ac-cepted as one of the most important aspects of the Industrial Engineering. The application and use of the simulation models has grown expo-nentially since the 50 until today. This is mainly because of the advances in the computation fi eld, but also because of the increased number (percentage) of acceptance by the academia and the industry (Robinson 2004a). The complexity of the simulated issues has been adapted to the complexity of the real world cases and has risen proportionally. Many of the tools and techniques used many years ago can not present the level of details that is needed today in some cases. One of the theses for future trends in the fi eld of simulation by Robinson (2004) is that in or-der to deal with this, a combination of techniques would be required. Also, in (Banks et al. 2003) few of the experts asked for bigger accent to be put in interoperability of simulation software. In that direction, the best from the selected tech-niques would be taken and they would comple-ment each other, resulting in the synergy factor.

135*Ss. Cyril and Methodi University, Faculty of Mecahical Engineering, Skopje, Macedonia; [email protected] Paper presented at the SIE 2012


In this paper, a comparison and combination of System Dynamics and Discrete Event Simula-tion (DES) will be presented. At the end one re-search example will be presented, showing why and when this should be done.

SYSTEM DYNAMICS

System Dynamics (SD) is a relatively new tech-nique that has been populated in the last 20 years. The basic principle underlying system dynamics is that the structure of a system deter-mines its behaviour over time (Forrester 1968; Sterman 2006). SD is all about the whole and looking at the system as a unit. In normal cases, a lot of people use the divide-and-conquer sys-tem in order to solve complex problems. The phi-losophy of SD is that every element is connected somehow with other element(s) and those rela-tionships determine how the system performs over time. It is best used when modelling very complex systems that are very hard to perceive and understand. There are two main approaches that help defi ne a SD model. The fi rst one is the causal loops (and feedback loops), which are widely spread and very useful. Most of the time, they are the fi rst step in developing a SD model, helping in the conceptualisation. The second tool is the stock and fl ow diagrams, which aid to describe the model using data. The easiest way to de-scribe this is to think of models like system of wa-ter tanks with pipes and valves.

In the research conducted by Helal et al. (2007) they have stated that using SD at the operation-al level of the manufacturing system has failed to offer the needed granularity (Godding et al., 2003; Barton et al., 2001; Baines and Harisson, 1999; Bauer et al., 1982) [03, 07, 08]. The same was observed by Choi et al. (2006) who could not use SD to model the performance of the in-dividual processes in a software development system. In (zgn & Barlas 2009) the authors needed to increase the values of some variables by tenfold in order for SD to capture them and for the model to make sense. In addition, while SD permits the study of the sta-bility of the system over the long range, the trends of behaviour that it generates do not indicate what specifi c actions to be made and at what values of the action parameters. Such specifi cations require more detailed considerations that SD does not seem to work with, while DES has been effective at.

DISCRETE EVENT SIMULATION

DES is a more widely established simulation technique (Banks et al. 2004). The system is modelled as a series of events, that is, instants in time when a state-change occurs, (Robinson 2004). The models are stochastic and generally represent a queuing system. From the beginning until now, the models are based on a specifi c code that manages the simulation.

At the beginning, DES was developed and used in the manufacturing sector. But, as the times have changed, so have the areas where DES has found its applicability (hospitals, public offi c-es, document management etc.) Still, the main advantages and principles have never changed no matter if the simulated entities are products, people, documents etc. (Law 2006; Banks 1998).

COMPARISON

The SD and DES are very different approaches when trying to model a situation and there are distinctive communities that follow each, respec-tively. Little bit inspired by the title of Sherwood (2002), the following comparison will be made in order to clarify some things. If a task of an-alysing a forest is given to these two types of modellers, the SD modellers will try to look at the forest from above, or from far away. They will look at the landscape, see how the trees are spread and grouped, analyse the types of trees etc. Meanwhile, the DES modellers will try to go in the forest and search in it, look at every tree as an entity, the leaves of the trees, the structure of the trees etc. Having this in mind, it was not very diffi cult to accept SD a technique for the attempt to model strategic decisions and use DES for the operational processes and decisions. Based on the work of Chahal & Eldabi (2008c) and Lane (2000) a meta-comparison of both approaches is shown in Table 1.

There are numerous articles that describe and compare these techniques, particularly. Maybe one of the fi rst attempts was done by Ruiz-Usa-no et al. (1996) and before that Crespo-Mrquez et al. (1993) concentrating on discrete vs. con-tinues systems. All of them give some kind of proposition or direction what technique is most suitable in which cases. Most of them (Brailsford & Hilton 2001; zgn & Barlas 2009; Sweetser 1999; Huang et al. 2004; Wakeland & Medina 2010) share the idea of the authors, presented

136

MSc Bojan Jovanovski - Combining system dynamics and discrete event simulations - overview of hybrid simulation models

, 230


earlier that SD is more suitable when modelling a system and analysing it as whole and DES when more details are needed for the better rep-resentation. The researches have been mainly focused on developing two same models in the different approaches and analysing and sharing the results (Robinson & Morecroft 2006; Cre-spo-Mrquez et al. 1993; Wakeland & Medina 2010; Johnson & Eberlein 2002). Tako & Rob-inson (2008) have gone a step further and have

analysed a model building process by fi ve SD and fi ve DES modellers on a same situation- a prison population problem. One of the detailed and structured comparison has been done by Chahal & Eldabi (2008), dividing the analysis in more than thirty categories and explaining every one of them. There are even researches that deal with the third possible option when simulating (e.g. a supply chain) - simulation with agents and compare that along the previous two (Owen et al. 2008).

DES SDProblem

Seeking to understand the imapct of randomness on the system

Aiming to understand the feedback within the system and its impact

ScopeOperational Strategy / Policy

SystemHigh level of detail that physically

represents the system (detail complexity)More macro level of detail that summarises the

system (dynamic complexity) Methodology

Process view Systems view Philosophy

Randomness Feedback

Table 1: Meta-comparison of two approaches

COMBINING TWO MODELLING TECHNIQUES

There are couple of examples where the idea of hybrid models has been taken and proved use-ful, especially combining SD and DES. They will be analysed according the area/industry for which the model was created, how the models are connected, to which level this was applied in the organization, are the models dependent\in-dependent and the format of the hybrid model. In the next section, we will share our insights re-garding each of these issues and present you an example of a hybrid model being developed in mean time.

Area/industry of applicationIn the manufacturing industry, there is a good ex-ample for modelling hierarchical production sys-tems (Venkateswaran et al. 2004; Venkateswaran & Son 2005). The authors are concentrated on the production and production related elements, and have developed a SD model for the long-term plans (developed by the Enterprise-level decision maker) and short-term plans (devel-oped by the Shop-level decision maker). In the

paper (Rabelo et al. 2005) the authors have also examined a manufacturing enterprise, where they used SD to simulate a fi nancial (reinvest-ment) policy and DES to simulate the production process of one machine. They have represented the number of machines in the SD model, so by multiplying this variable with the output of the DES process they can generate the production output of the enterprise. Based on the framework of (Helal et al. 2007), same has been tested and a hierarchical production model has been devel-oped (Pastrana et al. 2010).

In the recent decade, the healthcare manage-ment has been seen as a very interesting fi eld for the industrial engineers (the Institute of In-dustrial Engineers have clas-sifi ed Healthcare Management in the same im-portance as Lean & Six Sigma, Supply Chain Management, Ergonomics, Quality systems etc. and some universities have a special IE cur-riculum for Healthcare management, e.g. TU Eindhoven ). This interest has also been shown in using the simulation for tackling issues in the healthcare. Chahal and Eldabi (2008a) have distinguished three formats how

137


, 230


the models inside a hybrid mode can commu-nicate: Hierarchical, Process- Environment and Integrated format. Later they have suggested a framework for hybrid simulation in the healthcare (Chahal & Eldabi 2010). In the work of Brailsford et al. (2010) the authors used the hybrid mod-els to represent two case. The fi rst one is when the DES model simulates a process of a patient being examined with a whole confi guration of a hospital, while the SD simulates the community and how a specifi c disease would spread. In the second case, the DES was used to simulate op-erations of a contact centre, and SD to simulate demo-graphic changes of the population being examined.

The use of hybrid modelling has found its ap-plicability in the civil engineering as well (Pen a-Mora et al. 2008; SangHyun Lee et al. 2007; Alvanchi et al. 2009) dealing with problems that are more complex to be solved with independent simulation models or project management tools. One of the few advantages that the authors found with this approach is the benefi t of propos-als for improvement they got from the models. In the same direction as the previous two papers, Martin and Raffo (2001) have also suggested a hybrid approach in the software industry. They have worked on an issue that can be managed with project management software as well, but they argue that the benefi t of the hybrid simula-tion is the experimentation that can be done. The use of agent-based modelling and SD as hybrid architecture can be also adapted for the automo-tive industry (Kieckhafer et al. 2009).

Type of connectionCombining the two different models in one hy-brid one is one of the most important thing in this whole process. This defi nes also how the models will communicate, share data, behave at a certain time point etc. Back in the 1999 there were two papers that stress out the possibilities and the advantages when using HLA (High Lev-el Architecture) to combine two or more models (Schulze 1999; Davis & Moeller 1999). Some research done so far has employed this tool in order to combine their models (Venkateswaran et al. 2004; Rabelo et al. 2003; Alvanchi et al. 2009). Clearly, the benefi ts are enormous, but also the effort, time and the technicality when us-ing this approach. Some have used a more usual ways to do this, like Excel and Visual Basic for Applications (Brailsford et al. 2010). There are even cases where a specifi c research has been

conducted in order to defi ne a generic module in order for SD and DES models to communicate and function (Helal et al. 2007).

There are even examples where the modellers have a single software solution (Anylogic, ) and combined a DES model with dif-ferential equations (Marin et al. 2010). Maybe it is not as same as the rest of the cases, but is worth mentioning as an approach.

Scope of the hybrid modelIn this section we would like to address at what scope is the hybrid model applied inside one area/organization; whether the hybrid model is about whole organization, two different function-al areas inside organization, only one functional area etc. For example, the work of Brailsford et al. (2010) has two different cases, but both use DES to simulate inner situations (hospital and calling centre operations), while SD simulates very broad scenarios (whole community or pop-ulation demographics). In the case of (Martin & Raffo 2001) the model is a representation of a project being under away. Rabelo et al. (2005) have modelled two different functional areas SD for the decisions concerning allocation of the fi nancial resources (of the plants) and DES for operational decisions of the plant (number of machines, people etc.). In the case of (Ven-kateswaran et al. 2004), the whole hybrid model is about the production in the enterprise; SD for the aggregate-planning level and DES for de-tailed-scheduling level.

Dependent\independent models inside hybrid modelThe intention of the authors was to distinguish if the singular models inside the hybrid one are independent or dependent on each other. The idea was that maybe two different modellers can model their own model independently and then combine the model, which is thought of as very practical and less time consuming. This was very hard to distinguish during the research of the papers, because there is not so specifi c in-formation regarding this issue. The authors have made experiments by themselves regarding this and have successfully paired two independent models.

138


, 230


Type of hybrid model formatChahal and Eldabi (2008a) have distinguished three formats how the models inside a hybrid model can communicate: Hierarchical, Process - Environment and Integrated format.

The works of (Venkateswaran et al. 2004; Ra-belo et al. 2005; Rabelo et al. 2003; Pastrana et al. 2010) have a hierarchical model. (Brailsford et al. 2010) and (Martin & Raffo 2001) both deal with processes and how the environment deals with the changes that they bring. In (Brailsford et al. 2010) the authors argue that no one until now has achieved to develop a hybrid model by the Integrat-ed format, but given the progress of the develop-ment of hybrid models, the gap is getting narrower.

EXAMPLE / CASE

For the research that is going on right now, we are in a process of developing a hybrid model, based on the case of one production enterprise.

Figure 1: Structure of the hybrid model

This was not possible to be done in DES only environment, and when we experimented only with SD we did not get the needed detail level of the production.

Because of the nature of the situation, we are developing two separate models. One SD model that will represent the top management deci-sion about how many sales personal need to be (hired/fi red) and one DES model about the process of production of the products been sold. The models are of hierarchical format accord-ing the classifi cation of (Chahal & Eldabi 2008a) and aid each other so that the number of sales personnel is according the demand, but also ac-cording the production capacity (from the DES model). The connection was established using the built-in functions of the used software (Plant Simulation for DES and PowerSim for SD) and we used Excel as data storage media through the simulation runs. The functioning of the hybrid model is presented in Figure 1.

139


, 230


The model works in that way that the SD model runs and triggers the DES model (the produc-tion) and sends the information regarding the demand. After the production cycle is fi nished, it sends back to the SD model the number of pro-duced products. This information is received and taken in the SD model in order to calculate the possible sales that is one of the main inputs for determining the number of sales people (which was the initial goal of the simulation model).

CONCLUCSIONS

This paper summarizes and analyses different hybrid simulation models from selected papers. This is a relatively new area and only handful of research papers exist. Based on the papers and the authors view, the need for this kind of models is very justifi ed and will be even more important in the near future. In order to get the most appro-priate and convincing representation of the real world, the suitable modelling approach should be used. Because we try to simulate very com-plex scenarios, the need for hybrid simulation and modelling is inevitable. For our needs, the usage of System Dynamics and Discrete Event Simulation has been proven most suitable.

ACKNOWLEDGEMENTS

This research is supported by a Macedonian-Austrian research project titled as Joint simula-tion model for strategic decision support funded by both Governments.

REFERENCES

Alvanchi, A., Lee, S. & AbouRizk, S.M., 2009. Modeling Architecture for Hybrid System Dy-namics and Discrete Event Simulation. ASCE Conference Proceedings, 339(41020), p.131. Abu Gaben, M., Krevinac, S., Vujoevi, M.: Modelujui sistemi u optimizaciji, (2007) Jour-nal of Applied Engineering Science (Istraivanja i projektovanja za privredu), no. 18, p. 37-47 Baines T, Harrison D. 1999. An opportunity for sys-tem dynamics in manufacturing system modeling. Production Planning and Control 10(6): 542-552Banks, J. et al., 2004. Discrete-Event Sys-tem Simulation (4th Edition), Prentice Hall.Banks, J., Hugan, J. & Lendermann, P., 2003. The future of the simulation industry. In E. S. Chick, P. J. Snchez, D. Ferrin, and D. J. Morrice, ed. Proceedings of the 2003 Winter Simulation Conference. pp. 2033-2043.

1)

2)

3)

4)

5)

Banks, J. ed., 1998. Handbook of simulation, Wiley Online Library. Barton J, Love D, Taylor G. 2001. Evaluat-ing design implementation strategies using simulation. International Journal of Produc-tion Economics 72: 285-299 Bauer C, Whitehouse G, Brooks G. 1982. Computer simulation of production system: Phase I. Technical Report COE No. 82-83-1. The University of Central Florida, Orlando, FLBrailsford, S.C., Desai, S.M. & Viana, J., 2010. Towards the holy grail: combining sys-tem dynamics and discrete-event simulation in healthcare. In B. Johansson et al., eds. Proceedings of the 2010 Winter Simulation Conference. pp. 2293-2303.Brailsford, S.C. & Hilton, N., 2001. A compar-ison of discrete event simulation and system dynamics for modelling health care systems. Food in Canada, pp.1-17.Curovi, D., Vasi, B., Popovi, V., Curovi, N.:Ekspertsko planiranje proizvodnje, (2008) Journal of Applied Engineering Science (Istraivanja i projektovanja u privredi), no. 20, p.49-57 Chahal, K. & Eldabi, T., 2010. A generic framework for hybrid simulation in health-care. In Proceedings of the 28th Interna-tional Conference of the System Dynamics Society. System Dynamics Society.Chahal, K. & Eldabi, T., 2008a. Applicabil-ity of hybrid simulation to different modes of governance in UK healthcare. In S. J. Mason et al., eds. Proceedings of the 2008 Winter Simulation Conference. pp. 1469-1477.Chahal, K. & Eldabi, T., 2008b. System Dy-namics and Discrete Event Simulation: A Meta-Comparison. In the proccedings of UK Operational Reserach Society Simulation Workshop. pp. 189-197.Chahal, K. & Eldabi, T., 2008c. Which is more appropriate: A multiperspective comparison between System Dynamics and Discrete Event Simulation. In Proceedings of the Eu-ropean and Mediterranean Conference on Infor-mation Systems. Al Bustan Rotana Hotel, DubaiChoi K, Bae D, Kim T. 2006. An approach to a hybrid software process simulation using the DEVS formalism. Software Process: Im-provement and Practice 11(4): 373-383

6)

7)

8)

9)

10)

11)

12)

13)

14)

15)

16)

140


, 230


Crespo-Mrquez, A., Usano, R.R. & Aznar, R.D., 1993. Continuous and Discrete Simu-lation in a Production Planning System. A Comparative Study. In E. Zepeda & J. A. D. Machuca, eds. Proceedings of the 1993 In-ternational System Dynamics Conference. System Dynamics Society, p. 8p.Davis, W. & Moeller, G.L., 1999. The High Level Architecture: is there a better way? In P. A. Farrington et al., eds. Proceedings of the 1999 Winter Simulation Conference. pp. 1595-1601.Forrester, J.W., 1968. Principles of Systems, Pegasus Communications.Godding G, Sarjoughian H, Kempf K. 2003. Semiconductor supply network simulation. The Winter Simulation Conference, Dec 7-10, New Orleans, LAHelal, M. et al., 2007. A methodology for In-tegrating and Synchronizing the System Dy-namics and Discrete Event Simulation Para-digms. Industrial Engineering.Huang, P. et al., 2004. Utilizing simulation to evaluate business decisions in sense-and-respond systems. Simulation, (2000).Johnson, S. & Eberlein, B., 2002. Alternative modeling approaches: a case study in the oil & gas industry. In 20th System Dynamics Conference, Palermo, Italy.Kieckhafer, K. et al., 2009. Integrating agent-based simulation and system dynamics to support product strategy decisions in the auto-motive industry. Proceedings of the 2009 Win-ter Simulation Conference, pp.1433-1443.Lane, D. C., 2000. You Just Dont Under-stand Me: Modes of failure and success in the discourse between system dynamics and discrete event simulation.LSE OR Working Paper 00.34.Law, A., 2006. Simulation Modeling and Analysis, Mcgraw Hill Higher Education.Lee, SangHyun, Han, S. & Pen a-Mora, F., 2007. Hybrid System Dynamics and Discrete Event Simulation for Construction Manage-ment. Computing in Civil Engineering 2007, (May 2011), p.29.Marin, M. et al., 2010. Supply chain and hy-brid modeling: the panama canal operations and its salinity diffusion. In B. Johansson et al., eds. Proceedings of the 2010 Winter Simulation Conference. pp. 2023-2033.

17)

18)

19)

20)

21)

22)

23)

24)

25)

26)

27)

28)

Martin, R. & Raffo, D., 2001. Application of a hybrid process simulation model to a soft-ware development project. Journal of Sys-tems and Software, 59, pp.237-246.Meadows, D.H., 2008. Thinking in Systems: A Primer D. Wright, ed., Chelsea Green Publishing.Owen, C., Love, D. & Albores, P., 2008. Se-lection of simulation tools for improving supply chain performance. Business, pp.199-207.Pastrana, J. et al., 2010. Enterprise scheduling: Hybrid and hierarchical issues. In B. Johans-son et al., eds. Proceedings of the 2010 Winter Simulation Conference. IEEE, pp. 33503362.Pen a-Mora, F. et al., 2008. Strategic-Op-erational Construction Management: Hybrid System Dynamics and Discrete Event Ap-proach. Journal of Construction Engineering and Management, 134(9), p.701.Rabelo, L. et al., 2003. A Hybrid Approach to Manufacturing Enterprise Simulation. Pro-ceedings of the 2003 International Confer-ence on Machine Learning and Cybernetics; wintersim, 2, pp.1125-1133.Rabelo, L. et al., 2005. Enterprise simula-tion: a hybrid system approach. International Journal of Computer Integrated Manufactur-ing, 18(6), pp.498-508.Robinson, S., 2004a. Discrete-event simula-tion: from the pioneers to the present, what next? Journal of the Operational Research Society, 56(6), pp.619-629.Robinson, S., 2004b. Simulation: The Prac-tice of Model Development and Use, John Wiley& Sons Ltd.Robinson, S. & Morecroft, J., 2006. Compar-ing discrete-event simulation and system dy-namics: modelling a fi shery. In Proceedings of the Operational Research Society Simula-tion Workshop. pp. 137148.Ruiz-Usano, R. et al., 1996. System Dynam-ics and Discrete Simulation in a Constant Work-in-Process System: A Comparative Study. In G. P. Richardson & J. D. Sterman, eds. Proceedings of the 1996 International System Dynamics Conference. System Dy-namics Society, pp. 457-460.Schulze, T., 1999. On-line data processing in simulation models: new approaches and possibilities through HLA. In P. A. Farrington et al., eds. Proceedings of the 1999 Winter Simulation Conference. pp. 1602-1609.

29)

30)

31)

32)

33)

34)

35)

36)

37)

38)

39)

40)

141


, 230


Spasojevi, V., Klarin, M., Curovi, D.: Dimen-zije menadmenta kvalitetom isporuioca u industrijskim preduzeima Srbije, (2009), Journal of Applied Engineering Science (Istraivanja i projektovanja u privredi), no. 23-24, p. 67-70Sherwood, D., 2002. Seeing the Forest for the Trees: A Managers Guide to Applying Systems Thinking, Nicholas Brealey Publish-ing.Sterman, J.D., 2006. Business Dynamics, McGraw-Hill.Sweetser, A., 1999. A Comparison of System Dynamics ( SD ) and Discrete Event Simula-tion ( DES ). System, p.8.Tako, A.A. & Robinson, S., 2008. Model building in System Dynamics and Discrete-event Simulation: a quantitative comparison. Analysis.Venkateswaran, J. & Son, Y.J., 2005. Hybrid system dynamicdiscrete event simulation-based architecture for hierarchical production planning. International Journal of Production Research, 43(20), pp.4397-4429.

41)

42)

43)

44)

45)

46)

Venkateswaran, J., Son, Y.J. & Jones, A., 2004. Hierarchical production planning using a hybrid system dynamic-discrete event sim-ulation architecture. Proceedings of the 2004 Winter Simulation Conference, pp.1094-1102.Wakeland, W.W. & Medina, U.E., 2010. Comparing Discrete Simulation and System Dynamics: Modeling an Anti-insurgency In-fl uence Operation. Proceedings of the 28th International Conference of the System Dy-namics Society, (1991), pp.1-23.zgn, O. & Barlas, Y., 2009. Discrete vs. Continuous Simulation: When Does It Mat-ter? Proceedings of the 27th International Conference of The System Dynamics Soci-ety, (06), pp.1-22

47)

48)

49)


142


, 230

Paper number: 10(2012)3, 231, 143-146

FUZZY SYSTEMS TO SUPPORT INDUSTRIAL ENGINEERING MANAGEMENT

Dr Isabel L. Nunes *Universidade Nova de Lisboa, Faculty of Science and Technology, Portugal

doi:10.5937/jaes10-2510

* Universidade Nova de Lisboa, Faculty of Science and Technology, Portugal; [email protected] Paper presented at the SIE 2012

143

This paper presents the potentialities of Fuzzy Set Theory to deal with complex, incomplete and/or vague information which is characteristic of some industrial engineering problems. Two systems that were developed to support the activities of industrial engineering managers are presented as examples of the use of this mathematical methodology.Key words: Work related musculoskeletal disorders, Ergonomics, Resilience, Supply chain, disturbances, Industrial engineering, Fuzzy systems

INTRODUCTION

Many problems in Industrial Engineering are complex and have incomplete and/or vague in-formation. Also the dynamics of the decision en-vironment limit the specifi cation of model objec-tives, constraints and the precise measurement of model parameters (Kahraman et al., 2006). Fuzzy Set Theory (FST) developed almost fi fty years ago by L.A. Zadeh (Zadeh, 1965), is an excellent framework to help solve these prob-lems. According to (Kahraman, 2006) Industrial Engineering is one of the branches where FST found a wide application area. (Kahraman et al., 2006) present an extensive literature review and survey of FST in Industrial Engineering. A review of the application of FST to human-centred sys-tems can be found in (Nunes, 2010). This paper presents two application examples of fuzzy decision support systems aiming to sup-port industrial engineering managers in two dif-ferent areas of risk management: ergonomics and supply chain disturbances management.

FUZZY SET THEORY

FST provides the appropriate logical/mathemati-cal framework to deal with and represent knowl-edge and data, which are complex, imprecise, vague, incomplete and subjective (Zadeh, 1965). It allows the elicitation and encoding of imprecise knowledge, providing a mean for mathematical modeling of complex phenomena where tradi-tional mathematical models are not possible to apply.A fuzzy set (FS) is the generalization of classical (crisp) set. By contrast with classical sets which

present discrete borders, FS presents a bound-ary with a gradual contour. Formally, let U be the universe of discourse and u a generic element of U, a fuzzy subset A, defi ned in U, is a set of dual pairs:

where A(u) is designated as membership func-tion or membership grade u in A. The member-ship function associates to each element u, of U, a real number A(u), in the interval [0,1], which represents the degree of truth that u belongs to A.Using FST it is possible to evaluate the degree of membership of some observed data, originating either from an objective source or a subjective source, to some high-level concept. Let us con-sider, for example, the evaluation of the delay disturbance based on the continuous member-ship function presented in Figure . A low degree of membership to the disturbance concept (i.e., values close to 0) means the delay is accept-able; while a high degree of membership (i.e., values close to 1) means the delay is unaccept-able (Nunes & Cruz-Machado, 2012).The human-like thinking process, i.e., approxi-mate reasoning is well modeled using Fuzzy Logic (FL), which is a multi-value logic concept based on FST (Zadeh, 1996). Thus FL permits to process incomplete data and provide ap-proximate solutions to problems that cannot be solved by traditional methods. It allows handling the concept of partial truth, where the truth value may range between completely true and com-pletely false. Furthermore, when Linguistic Vari-ables (LV) are used, these degrees may be man-aged by membership functions (Zadeh, 1975a;


1975b; 1975c). A LV is a variable that admits as values words or sentences of a natural language (Figure 2), their terms can be modifi ed using linguistic hedges (modifi ers) applied to primary terms.

Figure 1: Fuzzy set delay disturbance (Nunes & Cruz-Machado, 2012)

FST can be used in the development of, for in-stance, fuzzy expert systems or fuzzy decision support systems. The following cases are exam-ples of these types of systems that can support industrial engineering managers activities.

EXAMPLES OF FUZZY SYSTEMSFAST ERGO X

Work-related musculoskeletal disorders (WMSD) are diseases related and/or aggravated by work that can affect the upper and the lower limbs as well as the neck and lower back areas. WMSD can be defi ned by impairments of bodily struc-tures such as muscles, joints, tendons, ligaments, nerves, bones and the localized blood circulation sys-tem, caused or aggravated primarily by work itself or by the work environment (Nunes & Bush, 2012).

Figure 2: Linguistic variable inadequacy used to evaluate protection inadequacy (Nunes & Simes-Marques, 2012)

FAST ERGO X (Figure 3) is a fuzzy expert sys-tem designed to identify, evaluate and control the risk factors existing in a work situation, due to lack of adequate ergonomics that can lead to the devel-opment of WMSD (Nunes, 2006; Nunes, 2009).Fast Ergo X evaluates the risk factors based on objective and subjective data and produce re-sults regarding the degree of possibility of de-velopment of WMSD on the upper body joints and about the main contributing risk factors. The results (Conclusions) are presented both quanti-tatively (as membership degrees to inadequacy fuzzy set, defi ned in the interval [0, 1]) and quali-tatively (as terms of a linguistic variable inten-sity). For instance The possibility for develop-ment of a WMSD on the Right Wrist is extreme (0.92). The Conclusions can be explained (Ex-planations) by presenting the computed risk fac-tors inadequacy degrees that contributed to the overall result, e.g. The number of Repetitions performed by the Right Wrist is very high. The system also presents Recommendations that users can adopt to eliminate or at least to reduce the risk factors present in the work situation. Some of the recommendations are in the form of good practices and graphical illustrations.

Figure 3: Activities performed on the analysis of a work situation by FAST ERGO X (Nunes, 2009)

A Fuzzy Decision Support System to manage supply chain disturbances

Supply Chains (SC) are subject to disturbanc-es that can result from acts or events that are originated inside of the SC (e.g., supplier failures, equipment breakdown, employees absenteeism) or may result from extrinsic events (e.g., social turmoil, terrorist attacks, or acts of God such as volcanic eruptions, hurricanes or earthquakes)

Dr Isabel L. Nunes - Fuzzy systems to support industrial engineering management

144 , 231

Journal of Applied Engineering Science 10(2012)3 145

(Nunes & Cruz-Machado, 2012). The Supply Chain Disturbance Management Fuzzy Deci-sion Support System (SCDM FDSS) developed by (Nunes et al., 2011) was designed to assess the SC and the organizations belonging to the SC based on their performance considering the following different scenarios, normal operation, when a disturbance occurs and when mitigation and/or contingency plans are implemented to counter the disturbance. The aim of the SCDM FDSS is to assist managers in their decision pro-cess related with the choice of the best opera-tional policy (e.g., adoption of mitigation and/or contingency plans) to counter disturbance effects that can compromise SC performance. The system combines the use of FST to model the uncertainty associated with the disturbances and their effects on the SC with the use of dis-crete-event simulations using the ARENA soft-ware (a commercial simulation tool) to study the behavior of the SC subject to disturbances, and the effects resulting from the implementation of mitiga-tion or contingency plans. The block diagram of the proposed SCDM FDSS is illustrated in Figure 4.

Figure 4: Relationship between SCDM FDSS and ARENA (adapted from (Nunes et al., 2011)).

The Inference Engine offers the reasoning ca-pability of the system. It performs the FDSS analysis using a Fuzzy Multiple Attribute Deci-sion Making model, and fuzzy data that charac-terizes the analyzed situation, using for instance fuzzifi ed Key Performance Indicators (KPI). The inference process includes 7 steps (Nunes et al., 2011):

Computing the KPI for each scenario and SC entity for each simulation time period. The KPI are obtained at the end of each ARENA SC simulation;Synthesizing the time discrete KPI into an

1.

2.

equivalent KPI for the relevant period con-sidered (obtained through a mean function);Fuzzifying the equivalent KPI into a fuzzy KPI (FKPI). Fuzzy sets convert KPI in normalized FKPI, i.e., fuzzy values in the interval [0, 1], where a fuzzy value close to 0 means a bad performance and a fuzzy value close to 1 means a good performance;Computing of a fuzzy performance Category Indicator (CI) for each scenario and SC en-tity using weighted aggregations of FKPI, through the following expression:

3.

4.

where:CIik is the fuzzy performance Category Indica-tor for ith category of KPI and for kth SC entity;FKPIijk is the j

th Fuzzy Key Performance Indi-cator of the ith category of KPI and the kth SC entity;wijk is the weight of j

th Fuzzy Key Performance Indicator of the ith category of KPI and the kth SC entity.

Computing of a fuzzy Performance Index (PI) for each scenario and SC entity using a weighted aggregation of CI, using the follow-ing expression:

5.

where:PIk is the Performance Index of k

th SC entity;CIik is the fuzzy performance Category Indica-tor for ith category of KPI and for kth SC entity;wijk is the weight of the ith category of KPI and the kth SC entity.

Computing of a fuzzy Supply Chain Perfor-mance Index (SCPI) for each scenario using a weighted aggregation of PI, using the fol-lowing expression:

6.

where:SCPI is the Supply Chain Performance Index of the SC for the current scenario;PIk is the Performance Index of k

th SC entity for the current scenario;wk is the weight of the k

th SC entity.Ranking alternatives. Scenario results for each entity and for the SC are ranked based on their PI and SCPI, respectively, in order to identify the operational policy with more merit.

7.


, 231


Using the results produced by the system (PI and SCPI) managers can: forecast the effects of disturbances in SC entities and on a SC as a whole; analyze the reduction of the negative impacts caused by the disturbance when op-erational policies are implemented; and select-ing the operational policy that makes the SC more resilient. The best operational policy cor-responds to the implementation that leads to the highest PI/SCPI value. The use of fuzzy modeling and simulation offers several benefi ts, inter alia, promotes a proactive SCDM, and improves the understanding of the impact of applying different operational policies meant to prevent or counter the effects of distur-bances, allowing the selection of the ones that are more effective and effi cient.

CONCLUSIONS

FST has been used since the sixties as a way to deal with complex, imprecise, uncertain and vague data in different areas of industrial engineering. In this paper the main characteristics and advan-tages of the use of FST were highlighted. Two examples of fuzzy systems applied to support decision-makers in the industrial engineering context were very briefl y presented (one in the fi eld of ergonomics and other in the fi eld of sup-ply chains management). The objective was to raise awareness to the in-dustrial engineers present in this conference to the potential that FST offers as a modelling tool to address complex phenomena that many in-dustrial problems present.

REFERENCES

Kahraman, C. (2006). Preface. In: Fuzzy Ap-plications in Industrial Engineering, C. Kah-raman (ed). Springer, New YorkKahraman, C., Glbay, M. & Kabak, . (2006). Applications of Fuzzy Sets in Indus-trial Engineering: A Topical Classifi cation In: Fuzzy Applications in Industrial Engineering, C. Kahraman (ed). pp. 1-55. Springer, New YorkNunes, I. L. (2006). ERGO_X - The Model of a Fuzzy Expert System for Workstation Er-gonomic Analysis. In: International Encyclo-pedia of Ergonomics and Human Factors, W. Karwowski (ed). pp. 3114-3121. CRC Press, ISBN 041530430X.

1)

2)

3)

Nunes, I. L. (2009). FAST ERGO_X a tool for ergonomic auditing and work-related mus-culoskeletal disorders prevention. WORK: A Journal of Prevention, Assessment, & Reha-bilitation, 34(2): pp. 133-148. Nunes, I. L. (2010). Handling Human-Cen-tered Systems Uncertainty Using Fuzzy Log-ics A Review. The Ergonomics Open Jour-nal, 3: pp. 38-48. Nunes, I. L. & Bush, P. M. (2012). Work-Re-lated Musculoskeletal Disorders Assessment and Prevention. In: Ergonomics - A Systems Approach, I. L. Nunes (ed). pp. 1-30. InTech, 978-953-51-0601-2.Nunes, I. L. & Cruz-Machado, V. (2012). A fuzzy expert system model to deal with sup-ply chain disturbances. Int. J. Decision Sci-ences, Risk and Management, 4(1/2): pp. 127151. Nuens, I. L., Figueira, S. & Machado, V. C. (2011). Evaluation of a Supply Chain Perfor-mance Using a Fuzzy Decision Support Sys-tem. Proceedings of The IEEE International Conference on Industrial Engineering and Engineering Management - IEEM2011Sin-gapore, 6-9 DecNunes, I. L. & Simes-Marques, M. (2012). Applications of Fuzzy Logic in Risk Assess-ment - The RA_X Case. In: Fuzzy Inference System theory and applications, M. F. Azeem (ed). pp. 22-40. InTech, 978-953-51-0525-1. Zadeh, L. A. (1965). Fuzzy sets. Information and Control, 8(3): pp. 338-353. Zadeh, L. A. (1975a). The concept of a lin-guistic variable and its application to approxi-mate reasoning-part I. Information Sciences, 8(3): pp. 199-249. Zadeh, L. A. (1975b). The concept of a lin-guistic variable and its application to approxi-mate reasoning-part II. Information Sciences, 8(4): pp. 301-357. Zadeh, L. A. (1975c). The concept of a lin-guistic variable and its application to approxi-mate reasoning-part III. Information Scienc-es, 9(1): pp. 43-80. Zadeh, L. A. (1996). Fuzzy Logic = Comput-ing with words. IEEE Transactions on Fuzzy Systems, 4(2): pp. 103-111.

4)

5)

6)

7)

8)

9)

10)

11)

12)

13)

14)



146 , 231

Paper number: 10(2012)3, 232, 147-152

A COMBINING GENETIC LEARNING ALGORITHM AND RISK MATRIX MODEL USING IN OPTIMAL

PRODUCTION PROGRAMDr Mirjana MisitaUniversity of Belgrade, Faculty of Mechanical Engineering, Belgrade, SerbiaDr Galal Senussia *Omar El-Mohktar University, Industrial Engineering Department, El-Baitha, LibyaMSc Marija MilovanoviUniversity of Belgrade, Faculty of Mechanical Engineering, Belgrade, Serbia

doi:10.5937/jaes10-2523

One of the important issues for any enterprises is the compromise optimal solution between inverse of multi objective functions. The prediction of the production cost and/or profi t per unit of a product and deal with two obverse functions at same time can be extremely diffi cult, especially if there is a lot of confl ict information about production parameters. But the most important is how much risk of this compromise solution. For this reason, the research intrduce and developed a strong and cab-able model of genatic algorithim combinding with risk mamagement mtrix to increase the quality of decisions as it is based on quantitive indicators, not on qualititive evaluation. Research results show that integration of genetic algorithim and risk mamagement matrix model has strong signifi cant in the decision making where it power and time to make the right decesion and improve the quality of the decision making as well.Key words: Multi-objective function, Genetic Algorithim, Risk Management, Optimum Production Program, Matrix, Costs

INTRODUCTION

The analysis of the production program of enter-prises is an important and complex segment of managing the enterprise, considering the fact that it infl uences all elements or resources, such as planning of the material, human resources, ma-chinery resources, research and development, marketing etc. All of these resources infl uence in multi-criteria optimization of production program. To reduce and improve the decesion making qual-ity, it is important and necessary to evaluate them to minimize the risk of operating losses.

In investigations carried out to date the produc-tion program optimization was based on multi-criteria approach using linear functions [01, 09]. Using nonlinear functions in multi-objective op-timization enables the application of genetic al-gorithms and is a step forward in the analysis of the product optimal quantities to maximize pro-duction resources utilization [06, 10, 07]. On the other hand, economic calculation of the product cost price is a complex procedure, so that the analysis of optimal production program most

commonly employed direct costs to determine the cost price and to defi ne the cost function. However, cost functions based only on product variable costs cannot provide real optimal prod-uct quantities but are more suitable for ranking products that should be given priority in manu-facturing. Introducing overhead costs in the func-tion of cost price is a complex calculation pro-cedure most often diffi cult to understand by the user in a concrete enterprise, considering that it is not easy to classify individual expenses. It is thought that in metalworking companies, roughly assessing, direct costs account for about 60% of total unit costs, while the share of overhead costs is 40% [03].

In business of enterprises, there are several cat-egories of risk: risk of equipment failure (estimat-ed in relation to human safety, to evironment, to business losses, ect.), risk management as a se-curity measure, fi nacial risk assessment in cases of loan approval, quality management risk, ect.Generally, Enterprise Risk Management is rela-tively new concept, Fraser and Simskins [05] dis-tinguish following risk categories: Shareholder

147* Faculty of Mechanical Engineering, Kraljice Marije 16, 11000 Belgrade; [email protected] Paper presented at the SIE 2012


value risk, Financial reporting risk, Governance risk, Customer and market risk, Operations risk, Innovation risk, Brand risk, Partnering risk, Com-munications risk.

Risk management consisit of strategic risk, op-erational risk, fi nancial risk and risk acceptance. Strategic risk deal with competition, market posi-tion and economic conditions. Operational risk

ERM Framework

Process stepsType of risk

Hazard Financial Operational StrategicEstablish ContextIdentify riskAnalyze / quantify risksIntegrate riskAssess / Prioritize RisksTreat / Exploit RisksMonitor & Review

Concerned with the daily operations, precisely, to the consequences of daily decisions made in the company. The fi nancial risks are related to relations with banks and stockholders, etc. The types of risk and process steps itroduced by Risk Management Committee 2003 [11].

Table 1: Enterprise Risk Management [8]

The risk is defi ned as product of probability and consequence of certain events, which can be ex-pressed in formula:

R = P*Q

P - Probability a particular event.Q Consequences of particular event.

For any enterprises, there are external and in-ternal of n-sources of risk. The total risk will rep-resented by high-risk, medium-risk and low-risk sources of operating losses.

Figure 1: Risk Impact/Probability Chart

The based approach of applying risk are risk identifi cation - what can affect the implementa-tion of production program, risk analysis - defi n-ing the probability of occurrence of that, and risk assessment - determining the consequences, expressed in the form of operating losses.The most low-risk sources of operating losses refer to good quality decision. Figure 2 shows the map for identifying Business risks.

Glover at all [9] states that the most real life opti-mization and scheduling problems are too com-plex to be solved completely and that the com-plexity of real life problems often exceeds the

148

Dr Mirjana Misita - A combining genetic learning algorithm and risk matrix model using in optimal production program

, 232


ability of classic methods. Miettinen [08] consid-ered that a key challenge in the real-life design is to simultaneously optimize different objectives through taking into account different criteria low cost, manufacturability, long life and good perfor-mance, which cannot be satisfi ed at the same time.

Profi t maximization is the main objective of busi-ness enterprises and as such the subject of nu-merous investigations. Profi t is defi ned as the difference between the total revenue generated by selling products on the market and the overall costs, i.e.:

P = TR TC

Where: P Total profi tTR Total revenue TC Total cost

When analyzing the possibilities of profi t maxi-mization, it is important to consider the fl uctua-tion of the TR and the TC. The TR depends on supply and market demands for particular types of goods, while the TC depends on different constraints faced by the company, such as the mechanical facilities, number and structure of employees, possibility of providing necessary specifi c materials for the manufacturing pro-cess implementation, delivery etc. For the com-pany, to be competitive on the market means to produce a product at an appropriate price and quantity with the use of capital and labor in the appropriate volume and costs. Therefore, profi t maximization refers to the optimization of vari-able parameters in the observed model, with given production constraints.

Where: P Profi tQ Quantity of productWpi Selling price of the ith productWvi Variable cost of the ith productTc Constant cost

In real life, the functions of dependence of pro-duction quantity and the TR and the TC are nonlinear. The maximum profi t is the maximum difference between the total profi t curve and the total cost curve, as represented in the Figure 3.

Figure 2: Graphic representation of profi t maximization

In real enterprises operating conditions the func-tions of the TR and the TC are nonlinear and to deter-mine them two different approaches must be applied.

The TR function consists of the sum of vari-able and fi xed costs, therefore, the sum of lin-ear mathematical form by applying the Lagrange interpolation polynomial based on the values of variable costs from the previous period. It is possible to determine the nonlinear function of fi xed costs in a Lagrange interpolation polyno-mial is, in our case, a function of production quan-tity P (Q) with (n-1) level if we have n data points on the value of costs from the previous period.

Where:

149


, 232


METHODOLOGY

Methodological steps in developing model for risk management integration methodology and GA is shown on Figure 3.

Figure 3: Steps in developing model for risk management integration methodology and GA

CASE STUDY

In the company engaged in manufacturing pre-cision measuring instruments, we have analyzed the available data and formed nonlinear functions of the TR and the TC for the three products:a) ClocksRevenue function:

Cost function:

b) Water meterRevenue function:

Cost function:

150


, 232


c) Gas meterRevenue function:

Cost function:

The functions of criteria for profi t maximization will have the form:

Risk Source Risk rating1st Q. 2010Risk

2nd Q. 2010Risk

3rd Q. 2010Operation cost. Low Medium MediumLabor cost Low Medium MediumLubricant cost Low Low LowRaw martial cost Medium High HighFixed cost Medium Medium Mediumcapital availability Medium Medium Mediumbusiness operations supply chain management Medium Medium Medium

information technology Medium High Highplanning Medium Medium Highreporting Low Medium Medium

Table 2: Evaluation of risk sources and determination of trend

Respectively:

Constraints:If we consider the production capacity as a key constraint in the production quantity of some products, temporarily ignoring the structure of demand for mentioned products on the market, the restrictions are:

***Employees and raw material in the observed company are not of limiting character.

The Pareto front and values of the functions f1 and Figure 1 are shown in Figure 4. Figure 4: The Pareto front of optimum solutionFrom the Pareto front diagram, it is evident that optimum solution for production quantity and profi t maximization under given constraints is a set [2312; 219; 944], where the maximum profi t is 5,950,340 RSD calculated as max (f1-f2).

After getting the optimum solution, the second step is Identify and analysis of risk sources for the observed optimum product program. In our case, we have focused on the internal resources only. Identifi cation, evaluations, and determina-tion of trend are shown in the table 2.

This fi gure 5 shows a two-dimension risk map. The vertical axis rep

jaes - vol.10 - no.3

Documents

dr jezdimir kneevi

dr milo kneevi

dr gradimir danon

dr nenad aji

dr miodrag zec

dr vlastimir dedovi

dr duan milutinovi

dr jozef aronov