DEVELOPMENTS TO GRAPMCAL MODELLING METHODS AND THEIR APPLICATION IN NUNUFACTURING SYSTEMS ANALYSIS AND DESIGN
Gary John Colquhoun
A thesis submittedin fulfilment of the requirementsof Liverpool John Moores University for the degreeof Doctor of Philosophy
April 1996 ACKNOWLEDGEMENTS
An enormousdebt of gratitude is owed to Ray Baines my supervisor, friend, mentor and advisor for the last nine years. I-Iis enthusiasm,unstinting support, incisive critique and encouragement were freely given at a stage in his own careerthat gave him little time for the luxury of research. Through his humour I have kept a senseof proportion andby his exampleI have learnt the need for comniitment.
To my wife lean, my backboneand inspiration through yearsof part thne study and for her support through the changesthat shook our fives ten yearsago - this is the last, I promise.
My thanks to the present Director of the School of Engineeringand Technology Managementat Liverpool John Moores University and the past headsof the Department of Mechanical, Marine and Production Engineeringwho have provided their supportthroughout the work.
I would alsolike to acknowledgethe contributionof CannonIndustries, Scholl Consumer Products andBSD SteelService Centres for supportingthe casestudies reported in the thesis. In particular SteveMon at BSD Leedsfor his enthusiasticsupport of the useof modelling. ABSTRACT
Systemsmodelling methods have evolved out of the need to understandand analysecomplex human/machine/objectsystems. This researchaims to contribute to the use of static graphical modelling methods as a meansof gaining insight, analysing,designing and developing strategies for changein manufacturingenterprise systems.
The research is introduced through a discussion of the factors underlying modelling in manufacturing enterprises. The successof IDEF methods in the US defence manufacturing industry has strongly influenced this work and their developmentis reviewed. Findings from a survey of published IDEF methods research are reported and an investigation of IODEFo modelling practice in the UK is accomplishedusing a survey of modelling practitioners. The surveys and a comparative analysis of a range of modelling methods provide a basis to contextualise IDEFo modelling and review the potential to expand its capability. Reference models for manufacturingsystems are analysedand an BDEForeference model casestudy is used to demonstratetheir capacity to provide significant improvementsin aspectsof the modelling process. An investigation of systemsarchitectures and frameworks provides a precursor to proposalsto combine IDEFo and IIDEF3 methods in the context of a manufacturing enterprise modelling framework.
A modelling approachis developedby combiningthe two methodsin a novel composite IDEFO- 3 method that has the capability to describe the behavioural characteristicsof manufacturing systems. The model syntax is refined and tested and an IIDEFO-3methodology is developed. The new method is validated through a casestudy in a batch manufacturingcompany currently undergoing re-organisation to expand its manufacturingcapability. IDEFO-3 is shown to be capable of capturing and representingthe perceptionsof domain experts, providing sufficient resolution for analysis and substantiallyinfluencing the design of an improved TO BE model. The research concludes that IDEFO-3 is able to represent features of systems hitherto not possible to describe using IDDEFoor IIDEF3 as individual methods, that it allows meaningful analysisand that it can effectively contribute to manufacturingenterprise systemsanalysis and design.
iii CONTENTS
ACFNOWLEDGEDvlENTS ii ...... ABSTRACT iH ...... CONTENTS iv ...... LIST OF FIGURES viii ...... LIST OF TABLES )d ...... NOMENCLATURE xii ...... CHAPTER 1 INTRODUCTION 1.1 PREAMBLE 1 ...... 1.2 M[ANUFACTURING GLOBAL CONTEXT 2 -A ...... 1.3 SYSTEMS CONCEPTS 4 ...... 1.3.1 Manufacturing Systems 6 ...... 1.3.2 System 7 complexity and modelling...... 1.4 IDEFo FUNCTIONAL MODELLING 9 ...... 1.5 RESEARCH OBJECTIVES 10 ...... 1.6 RESEARCH M[ETHODOLOGY 11 ...... 1.7 CONTRIBUTION TO KNOWLEDGE 12 ...... 1.8 VALIDATION OF TIHE RESEARCH 13 ...... CHAPTER 2 THE IDEF METHODS 2.1 INTRODUCTION 15 ...... 2.2 A REVIEW OF TIiE ICAM PROGRAMME 15 ...... 2.3 A FUSTORICAL PERSPECTIVE OF THE IDEF METHODS 18 ..... 2.3.1 A development IDEFo 18 reviewof tile of ...... 2.3.2 The development IDEF 20 of associated methods...... 2.4 THE LATEST GENERATION OF IDEF METHODS 22 ...... CHAPTER 3 PUBLISHED IDEFo AND IDEF3 RESEARCH 3.1 INTRODUCTION 25 ...... 3.2 STANDARD REFERENCE SOURCES 26 ...... 3.3 A STATE OF THE ART REVIEW OF IDEFo 27 ...... 3.3.1 Category1: Descriptions basic 28 andreviews of principles...... 3.3.2 Category2: Performance 29 evaluations,and comparisons ...... 3.3.3 Category3: Proposed 32 enhancementsof the method ...... 3.3.4 Category4: Applications 36 of themethod ...... 3.4 A REVIEW OF PUBLISBED IDEF3 RESEARCH 37 ...... 3.5 A SURVEY OF IDEFo PRACTITIONERS 40 ......
iv 3.5.1 Questionnairedesign 40 ...... 3.5.2 Results 41 of the survey...... 3.5.3 An 46 analysisof surveyresults ...... 3.6 CONCLUSIONS 49 ......
CHAPTER 4A COMPARATIVE ANALYSIS OF MANUFACTURING ENTERPRISE SYSTEMS ANALYSIS AND DESIGN METHODS
4.1 INTRODUCTION 52 ...... 4.2 CONTEXT FOR A COMPARATIVE ANALYSIS 53 ...... 4.3 SOFT SYSTEMS MODELLING METHODS 55 ...... 4.4 INFORMATION SYSTEM MODELLING METHODS 57 ...... 4.5 'IS'MODELLING METHODS IN THE CONTEXT OF MANUFACTURING SYSTEMS 62 ...... 4.6 'HARD' SYSTEMS MODELLING METHODS 63 ...... 4.7 GRAPHICAL REPRESENTATION 64 ...... 4.8 A STRUCTURED COMPARATIVE ANALYSIS 65 ...... 4.9 CONCLUSIONS 71 ...... CHAPTER5 REFERENCESTRUCTURESFOR NL4NUFACTURING ENTERPRISE MODELLING 5.1 I1,4TRODUCTION 74 ...... 5.2 MANUFACTURING SYSTEMS ARCI-HTECTURES 74 ...... 5.2.1 CIM-OSA 76 ...... 5.2.2 TheZachman Framework 78 ...... 5.3 REFERENCE MODELS FOR KANUFACTURING SYSTEMS 82 ...... 5.3.1 ne applicationof an IDEForeference model of process plannin ...... 82 5.3.2 Derivingan AS-IS modelusing a genericmodel of process plannin ...... 85 5.3.3 Conclusions the 86 on useof a referencemodel ...... 5.4 A MANUFACTURING ENTERPRISE MODELLING FRAMEWORK 90 ...... 5.4.1 Conclusionsfrom themanufacturing enterprise modelling - framework ...... 93 CHAPTER 6A COMPOSITE BEHAVIOURAL MODELLING APPROACH 6.1 INTRODUCTION 102 ...... 6.2 A BLUEPRINT FOR ENTERPRISE MODELLING METHODS. 103 6.3 STAGE I-A CASE STUDY USING IDEFo AND IDEF3 105 ...... 6.3.1 A 1 110 reviewof stage ...... A3.1 IDEFO 224 ...... A3.2 IDEF3 ...... 226 A3.3 IDEFo AND IDEFix MODELLING TOOLS ...... 227 A3.4 IDEF1 ...... 231 A3.5 IDEFix 235 ...... A3.6 IDEF2 ...... 241 A3.7 IDEF4 ...... 247 APPENDIX 4 MODELLING METHODSREVIEW 253 ...... A4.1 THE CBECKLAND SOFr SYSTEMS METHODOLOGY 254 ...... A4.2 THE VIABLE SYSTEMS MODEL (VSM) 256 ...... A4.3 THE GRAI METHODOLOGY ...... 257 A4.4 THE STRUCTURED SYSTEMS ANALYSIS AND DESIGN METHODOLOGY (SSADM) ...... 259 A4.5 THE NUSSEN INFORMATION ANALYSIS METHOD (NIAM) 260 ...... A4.6 THE CONTROLLED REQUIREMENTS EXPRESSION (CORE) METHOD ...... 261 A4.7 DISCRETE EVENT SIMULATION 264 ...... A4.8 PETRINETS ...... 265 A4.9 CRITICAL PATH ANALYSIS (CPA) 266 ...... APPENDIX 5 EPSRC-CDPGRANT ANNOUNCEMENT 267 ...... APPENDIX 6 KBSI RESEARCH GRANT AGREEMENT 270 ...... APPENDIX 7 IDEF3 MODEL OF'RUN ZENO ...... 272 APPENDIX 8 THE MODELLING METHODS QUESTIONNAIRE 279 ...... APPENDIX 9 PROCESSPLANNING FOR BATCH M[ANUFACTURE 281 ...... APPENDIX 10 PART NUMBER/PROCESS MATRIX 284 ...... APPENDIX 11 BSD SC]HEDULING/MANAGEMENT SOFTWARE SPECIFICATION ...... 286 APPENDIX 12 ABSTRACTS FROM PAPERS PUBLISHED IN THE COURSE OF THE RESEARCH ...... 287
vii 7.4 Node Al I Establish 136 plate, capacityand componentrequirements ...... 7.5 Node Al 13 Process 137 contract orders...... 7.6 Node A3 Run 138 contract manufacture...... 7.7 Node A03 32 Produce to lists 139 - work ...... 7.8 Node A03 33 ManufactureKits 140 - ...... 7.9 Node A03 325 Prioritise 141 - components...... 7.10 An extract from 146 an analysisof alternativeprocess routes ...... 7.11 Node tree of IDEFO-3 model 'Run spot and Contract manufacturing 146 7.12 Node tree for IDEFO-3 model 'Run CEME centre...... 150 7.13 Node index for 'Run CEME centre...... 152 7.14 Node A-0 Run CEME 155 centre...... 7.15 Node AO Run CEME 156 centre...... 7.16 Node Al Manage 157 centre...... 7.17 Node A2 Plan Manufacturing 158 ...... 7.18 Node A21 Carry 159 out manufacturingplanning ...... 7.19 Node A22 Establish 160 manufacturingschedules ...... 7.20 Node A3 ManufactureKits 161 ...... 7.21 Node FEO A22F Establish 162 manufacturingschedules ...... 7.22 Node A03-221 Establish data 163 nesting ...... 7.23 Node A03-223 Control 164 work centre schedules...... APPENDIX 2 A2.1 IDEFo decomposition 205 ...... A2.2 IDEFO to PFN decomposition 206 ...... A2.3 IDEFO PFN 208 parent and child ...... A2.4 AN IDEFO-3 diagram 209 ...... A2.5 Types of terminating 210 processes...... A2.6 11E)EFO-3 syntax ...... 212 A2.7 The IDEFO-3 Methodology 218 ...... A2.8 The manufacturing enterprise framework 219 ......
APPENDIX 3 AM A ProcessFlow Network Run ZENO 226 - ...... A3.2 IDEF1 graphic notations to specify 233 relation classes...... A3.3 A simple IDEF, entity class diagram 233 ...... A3.4 A simple IDEF, entity class diagram showing 234 attribute classes...... A3.5 Graphical representation of IDEF, 236 x entities ...... A3.6 IDEF, relationship x cardinality ...... 237 A3.7 An IDEFx non-identifying relationship 238 ...... A3.8 A zero or one-to-something IDEF, non-identifying 238 x relationship ...... A3.9 IDEFx categorisation relationships 239 ...... A3.10 Extract from ' Engineering Data Control (EDC) in capital plant manufacture ...... 240 A3.11 An IDEF2 entity flow diagram 242 ...... A3.12 Alternative flows 243 entity ...... A3.13 A resource disposition tree 244 ...... A3.14 A System Control Network 245 ...... A3.15 The structure of an IDEF4 249 model ...... A3.16 The evolution of IDEF4 251 artefacts ......
ix APPENDIX 4 A4.1 The Viable SystemsModel 256 ...... A4.2 A GRAI for 'make 258 grid a to order' company...... A4.3 A GRAI Net 259 ...... A4.4 A CORE tabular diagram 262 entry ...... A4.5 An individual diagram for 'Prove 263 thread the program' actioit ...... A4.6 Examples CORE box 264 of action syntax......
APPENDIX 7 A7.1 IDEF3 diagram 'Run ZENO ...... 273 A7.2 IDEF3 diagram 'Inspect and pack product...... 274 A7.3 IDEF3 diagram 'Monitor process...... 275 A7.4 IDEF3 diagram'Run 400 press...... 276 A7.5 IDEF3 diagram 'Set up 400 press...... 277 A7.6 IDEF3 diagram 'Carry out batch run...... 278
x LIST OF TABLES
Principal 41 CHAPTER3 3.1 activities againstnumber of respondents...... Primaryjob functions 42 3.2 againstnumbers of respondents...... 3.3 A Modelling 43 method/Applicationmatrix ...... 3.4 The IDEFO 44 subjectand size of models...... 3.5 The relationshipbetween primary job function and the reasonsfor IIDEFo 45 selecting ...... 3.6 A gradeduser view of the benefitsof someaspects of IDEFo 46 modelling...... 3.7 A gradeduser view of the benefits of someaspects of using the Design/IDEF tool 46 ...... Comparative 68-70 CHAPTER4 4.1ab, c analysisof modelling method characteristics......
APPENDIX 3 AM Examples IDEF4 terms 250 of ...... A3.2 Potential relationshipsbetween IDEF4 version 2 and other IDEF 251 methods......
xi NOAMNCLATURE
ABLE PM AutomatedBusiness Logic Engineering- ProcessModefling AFCAM Air Force ComputerAided Manufacturing AMTeC AdvancedManufacturing Technology Centre APEC Asia-PacificEconomic Co-operationorganisation APT Automatically ProgrammedTool BDF BusinessDesign Facility CASE tool (Texas Instruments) BM BehaviourModel (a sub-modelof IDEF4) BOM Bill Of Material BPR BusinessProcess Re-engmeenng BPwin IDEFIx modelling software tool (Logic Works Inc.) BSD British SteelDistribution CAD Computer Aided Design CAM-I ComputerAided Manufacturing - International CAPP ComputerAided ProcessPlanning CASE Computer Aided Software Engineering CATWOE Customer,Actor, Transformation,World-view, Owner, Environment CEME ConstructionEarth Moving Equipment centre CIM Computer IntegratedManufacturing CIM-OSA CIM - Open SystemsArchitecture CNC ComputerNumerical Control CORE The COntrolled RequirementsExpression method COSM0 Interactive Enterpriseengineering and CASE software tool CPA Critical Path Analysis CPN Coloured Petri-Net DACOM D. Appleton Company DFD Data Flow Diagram Model (a IDEF4) .DM Dynamic sub-modelof E/R Entity Relationship(model) EDC EngineeringData Control EPSRC-CDP Engineeringand Physical SciencesResearch Council-Control Design and Production group ER:win/ERX IDEF, x modellingtool ESPRIT European Strategic Programme for Research and Development in Information Technology FEDD For Early DomesticDissemination
xii FEO For Exposition Only FIPS FederalInformation ProcessingStandard FMEA Failure Mode Effect Analysis GDP GrossDomestic Product GERTE GraphicalEvaluation and Review TEchnique GRAI Graphe a Rdsultats et Activit6s Interli6s (Graphs with Results and Activities Interrelated). HEPO IIierarchical Input-Process-Output HOS 11igherOrder Software ICAM IntegratedComputer Aided Manufacturingprogramme ICOM input, Output, Control, Mechanism(in IDEFo diagrams) IDEF The I-CAM DEFinition techniques(as definedby the ICAM programme) IDEF Integration DEFinition methods (as defined by KBSI, i. e. post-IDEF2 methods) . IDEFo The I-C AM DEFinition functional modelling method. IDEFO-3 The Compositebehavioural modelling method IDEFO-TD The IDEFo. Triple Diagonal technique IDEF, The I-CAM DEFinition information modelling method IDEFIx The I-CAM DEFinition IDEF, eXtendeddata modelling method IDEF2 The I-CAM DEFinition dynamicsmodelling method IDEF3 Integration DEFinition - ProcessDescription capture method design IDEF4 Integration DEFinition - Object oriented modelling IDEF5 Integration DEFinition - Ontology descriptioncapture methods IDEF6 Integration DEFinition - Design rationalemodelling IDEF8 Integration DEFinition - User interfacemodelling IDEF9 Integration DEFinition - Scenario-drivenIS design IDEF10 integration DEFinition - Implementationarchitecture modelling IDEF12 Integration DEFinition - Organisationmodelling IDEF13 Integration DEFinition - Three- schemamapping design IDEF14 Integration DEFinition - Network design IDEM IntegratedDesign and Modelling methodology IEEE The Institute of Electrical and ElectronicsEngineers IFEM IntegratedFramework for EnterpriseModelling 11CE Information Integration for ConcurrentEngineering programme IISS IntegratedInformation Support System IMP integrated Manufacturing systemsdesign Procedure is Information System ISO International StandardsOrganisation KBSI Knowledge BasedSystems Inc.
xiii LDDT Logical DatabaseDesign Technique LOT Lexical Object Types (from the Nijssen Information AnalysisMethod) LUM_TM Liverpool University Method for Tool Management MAP ManufacturingAutomation Protocol MOSES Model Oriented SimultaneousEngineering System NCMS National Centrefor Manufacturing Science NET_TRANS Net Translator NIAM Nijssen Information AnalysisMethod NIST National Institute of Standardsand Technology NOLOT Non-Lexical Object Types (from NIAM) OR OperationsResearch 01ýff Object Method Technique OOA/OOD Object OrientedAnalysis and Design technique OSTN Object StateTransition Network (from IDEF3) PERT ProgrammeEvaluation and Review Technique PFN ProcessFlow Network (from IDEF3) ProSim IDEF3 softwaretool (KBSI) QC Quality Control RAM RandomAccess Memory SA StructuredAnalysis SE StructuredEngineering SADT Structured Analysisand Design Technique SAINT SystemAnalysis of IntegratedNetworks of Tasks modelling method SEE Society for EnterpriseEngineering SIMCON/2 Simulationcell Controller Sm Static model -a sub-modelof IDEF4 SSADM Structured SystemAnalysis Design Methodology SSM Soft SystemsMethodology. - Checkland TIF Toronto Industrial Fabrications UOB Unit Of Behaviour (from IDEF3) VSM The Viable SystemsModel WIP Work In Progress WITNESS A simulationsoftware tool (AT&T ISTEL) ZENO A manufacturingcell name(at Scholl ConsumerProducts)
xiv CHAPTER I
INTRODUCTION
1.1 PREAMBLE
The work presentedin this thesis grew from the early researchof Baines (Baines and Betts, 1984,Hughes and Baines, 1985) and from work carried out by the Advanced Manufacturing Technology Centre' (AMTeC) on the use of IIDEF methods in modelling manufacturing enterprises. These two threadswere the foundation of an AMTeC supported project that used modelling methodsto examinethe role of Computer Aided ProcessPlanning in a CIM environment (Colquhoun, 1988), findings from the work were subsequently published (Colquhoun, Gamble and Baines, 1989) and led to the line of research presented in this thesis.
This chapter discussesthe global businessenvironment facing UK manufacturing in the late 20P' century and presentsa casefor the adoption of modelling methods as tools for manufacturing enterprise analysis,planning and management. The key concepts that underpin enterprise modelling are discussedand the IIDEFomodelling method, that has substantially influenced the research, is introduced. The objectives and research methodology are defined and the chapter ends by reviewing the contribution to knowledge and validation of the work.
1AMTeC was an arm of the Machine Tool Industries ResearchAssociation, Macclesfield, Cheshire, LTK
I 1.2 MANUFACTURING -A GLOBAL CONTEXT
Manufacturing industry is undergoing a period of fundamental change. National and political boundaries are no longer limits to company organisation, trade blocs working to pool manufacturing ability and protect markets are emerging to change the nature of industrial co-operation and competition. Global communication links are forming new economic patterns sensitiveto industrial performancearound the world The pace of new technologiesand the rate at which new products enter the global marketplaceare all forces that make the manufacturing future for industrialised nations less predictable and more changeablethan at any previous period in the twentieth century.
For the UK and other industrialisedeconomies manufacturing is a key to national prosperity, 'In the UK manufacturingcontributes 22% of GDP, employsfive million people (2 1% of the workforce), generatesemployment for a further five million' (Deasley, and Collins, 1994) and, in 1993,82% of visible exports consistedof manufacturedgoods (FIMSO, 1995). This is despite the period between 1980 and 1992 when high exchangerates and a world-wide recessionforced the closure of many manufacturingenterprises accelerating the decline of the UK's industrial base and when Britain had its first manufacturingtrade deficit since the Industrial Revolution (Elliott and Beavis, 1994). In the UK now (1995) there is some cause for optimism in manufacturing, modest growth, lower unemployment and low inflation. There is, however, still concern about the size of its manufacturingbase, the loss of skills, the short term view taken by investors,the dependenceon inward investment (40% of total inward investment coming into the EU has been taken by the UY, (Elliott, 1995)) and the future for foreign owned enterprisesbased on -assemblyplants. Initiatives -such-as the Manufacturing and Construction Alliance (a parliamentarylobby for manufacturing chaired by Nicholas Wmterton W) have failed to persuadeGovernment, Banks, Pension funds or Venture Capitalists of the need to treat manufacturingdifferently and to finance renewal of manufacturingindustry through investmentin researchand development. Small and medium (SMNMs) for future in LIK manufacturing enterprises are critical the of manufacturing, the - 99% of VAT registered manufacturing units are SMXffis employing 45% of the manufacturingworkforce and producing 30% of total manufacturing output (Levy, 1993). International competitors such as Germany offer funding through a dedicated low interest,
2 governmentsupported bank for small enterprisesand in Japanaccess to new technology and long term finance for small enterprisesis supportedby the 'Small and Medium Enterprise Law'.
In the USA public expenditure has supported military development and its defence establishmenthas been 'the largest single source of R&D funding in the world' (Samuels, 1994). The USA is still the foremost industrial nation in the world today however their decline through a contracting manufacturingbase and competition in home and European markets from Japan and emerging Asian nations is recognised. They have respondedto changing global markets in the formation, in 1989, of the Asia-Pacific Economic Co- operation organisation (APEC) an ominous sign for Europe and the fragile UK manufacturing recovery. 'By 2000, APEC will be larger than the G-7 industrialised countries and will dominateUS trade' (Tran, 1993). -A marginalisationof manufacturingin the UK and Europe will result from a failure to develop existing markets and penetrate emergingmarkets in the Pacific rim andEastern Europe.
The: situation of UK companiesmeans it has never been more important for them to understandthe way they work and to adapt and respond to competition using innovative manufacturingand organisation. For manufacturerswithout flexibility or the ability to adapt, the new world is a major threat. In July 1994 a UK government research initiative (The Innovative Manufacturing Initiative (RQ) was launchedin recognition of the importance of longer term researchin manufacturingto the economic vitality of the UK. Processbased manufacturing is the first industrial sector to be targeted and will receive 120M research funding over the next ten years. An aim of the initiative is that by 2005 the UK will be a world leader in the modelling and analysisof the global supply, distribution and marketing of process products and will be the recognisedleader in businessprocesses for the process industries (EPSRC, 1994). Global competition, in a world where developedcountries are all capable of buying technology and skills, demands an ability to continually review an enterprise's operation for it to be able to maintain its position and gain advantage. Success may he in the exploitation of a unique product or technologybut. for most manufacturersit is more likely to be gained from what Drucker (1991) describesas 'working smarter', that is a
3 combination of technology, organisation, management,the control of information and an ability to continually adapt to new situations.
Most UK enterprisesmust meet the challenge of global competition with 'organisational structures that came of age in a different competitive environment' (Hammer, 1990) using existing plants, and cultures and often with less staff and with current practicesthat almost always constrain future development. Jones (1988) advocates that change should be a continuous process, 'managementis not about preservation of the status quo, it is about maintainingthe highestrate of changean organisationand the people can stand'. Gould and Dent (1993) propose that 'To be successfulmanufacturing companies will need a top down integration strategy covering all aspects of organisati0n, culture, people and business processesas well as the enablingtechnologies. Changeprocesses are likely to be complex, they may involve the whole enterprise and may challenge entrenched organisational structures, practices and culture. Understandingthe constraints to change that critically effect system efficiency and effectivenessis essentialto be able to organisethe analysisand planning necessarybefore changesare made. Those involved in implementing change are faced with a range of problems,processes are interdependent,important operational details are hidden, information systemsare complex and the consequencesof decisions delayed. The initial task is one of reducing the size of the problem to expose important details and describing behaviour to such an extent that it is possible to predict the consequencesof decisions. One strategy for the successful managementof change for re-engineering manufacturingenterprises is to employ practical, accessible,proven approachesto modelling that systemsengineers can use routinely for the analysisand design of complex established human/machinelobjectsystems. This thesis addressesresearch that is required.into the use of modelling methodsas tools to implementsuch a strategy..
1.3- SYSTEMS CONCEPTS
Rational-analyticalthinIdng and systemsthinIdng are two approachesthat have been applied to the study and analysis of manufacturing organisations. Rational-analytical thinking developedin sciencein the eighteenthcentury and, until the middle of the twentieth century,
4 was the acceptedphilosophy underlying problem solving in both scienceand social sciences. It is an approachto problem solving basedon three perspectives: (i) A reductionist view to deal with complexity, breakingdown problemsinto smallersimpler elementsto a level where the elementscan be understoodand the source of the problem exposed. (H) A positivistic view that arguesthat all the information about a problem can be known and unambiguous. (iii) A deterministic view that argues that the operation of a system can be predicted if system parameters are known and if they can be measured to identify changes. The philosophy is widely and successfullyapplied in mechanicalsystems for examplebut is less for dynamic human/machine/object suitable the complex, , systems that comprise manufacturingenterprises.
Alternatively, Systemsthinking approachestake a holistic view of a problem where no detail can be studied without knowing its relevanceto, or dependenceon, the larger system in which it exists. In systemsthinking the overall purposeof a systemis consideredcritical for investigation and, in contrast to rational-analytical thinking, subjectivity is an acceptable aspect of the approach. A fundamental aspect of systems thinking is the concept of $emergentproperties', which are properties a system possessesat one level becauseit is structured in a-certain way. The sameproperties may not be apparent in. sub-systemsat lower levels. The origins of Systemsthinking He in the roots of Western and Eastern culture. In 300 BC Plato discussedsystems of government and the need for a common purpose to unite the efforts of all the elementsof a system. The philosophies of Kant and Hegel in the nineteenth century introduced phenomenologyand 'inquiring systems' laying the foundation for the 'learning organisation'. However recognition of the importance of Systemsconcepts to deal with complexity in engineering,social scienceand managementare relatively recent. In 1956 the biologist Ludwig von Bertalanffy (Bertalanfly, 1956) published his General SystemsTheory and predicted that 'the concept of a 'system' is to become a fidcrum in modem scientific thought' Nfiller (1978) defines the General Systems Theory as a 'set of related definitions, assumptionsand propositions which deal with reality as an integrated hierarchy of organisations of matter and energy.' Bertalanfly's work advocatedthe needto study a systemas a entity and to open interactions for examinationin contrast to the then contemporary approach 'to isolate phenomenain narrowly confined contexts' the rational analyticalapproach.
5 In describing the, then new, systems approach to scientific thought Ackoff (1959) propounded 'these. research pursuits and many others are being interwoven into a co- operative researcheffort involving an ever widening spectrum of scientific and engineering disciplines. We are participating in what is the most comprehensiveeffort to attain a synthesis of scientific knowledge yet made'. In the same year Emery and Trist (1959) recognisedthat 'The analysisof the characteristicsof enterprisesas systemswould appearto have strategic significancefor furthering our understandingof a great number of specific industrial problems' and that 'the more we know about these systemsthe more we are able to identify what is relevant to a particular problem and to detect problems that tend to be missed by the conventional framework of problem analysis.' Bertalanffy (1962) saw the importance of systems thinking for manufacturing and proposed that 'entities whose componentsare most heterogeneous- men, machines,buildings, monetary and other values, inflow of raw material, outflow of product and many other items can successfidlybe submitted to systemsanalysis' Problem solving using systemsthinking has evolved into diverse streams,Beer's organisationalcybernetics approach (Beer, 1979), Checkland's soft systemsthinking (Checkland, 1981) and the SystemsAnalysis movement for example, in different spheresbut sharingcommon concepts and terminology. The word 'system' is now commonly used to describea wide variety of organisedcollections of things, functions, or people. A filing system,a penal system,a control systemfor exampleall reflect the Oxford Dictionary definition of a system as a: 'complex whole, set of connected things or parts, organisedbody of material or immaterialthings. '
1.3.1 Manufacturing systems
The focus of this work is on manufacturingenterprise systems which comprise of elements i. e. humans,equipment, machines, buildings, objects or material (which can be describedby a noun or noun phrase) related and organisedas processesto interact with each other and their environment with the objective of manufacturing products to satisfy consumer demands.
6 The concepts of manufacturing systems described by Hitomi in 1990 as 'integrated manufacturing unifying material flow (manufacturing processes) and information flow (production management)'are expandedin his exhaustivereview of their developmentin 1994 (Hitomi, 1994). He that based three (i) . proposes manufacturingsystems are on aspects handling, (H) structure -a unified assemblyof workers, production facilities and material (iii) transformation - material flow and the conversion of raw material to products, procedure - the managementcycle, planning, implementation and control. For those involved, the needto understand,design and develop such systemshas given rise to research aimed at providing suitable methods and tools to deal with their often complex and ill- defined nature.
1.3.2. System complexity and modelling
Measures of system comple)dtyare difficult to establish,Klir (1985) arguesthat there has been 'virtually no sufficiently comprehensivestudy that is oriented to capturing its general characterisation(i. e. measuresof complexity). The Oxford English dictionary describes complexity as 'hard to unravel' and Webster's English dictionary as 'marked by an involvement of many parts, aspects or notions, and necessitating earnest study or examinationto understandor cope with'. Systemcomplexity is more generally describedby Flood and Carson, (1988) and others as a relative term involving (i) a number of elements, (H) a number of relationshipsbetween elementsand (iii) a perception or viewpoint. Given that a measureof complexity involves a perception of a systemit must also implicitly involve the ability of an observerto understandthe system. Klir (1985) concludesthat 'complexity is in the eyes of the observer' and cites Ashby's (1973) incisive definition of a measureof systemscomplexity as 'the quantity of information required to describethe system'. Baines (1984) consideredmodels to assessthe complexity of the production control function in a manufacturing system, stressingthat three properties are of primary importance: (i) The model must provide managerswith an easyto use scientific approachfor assessment.(H) It must be able to deal with an extremely variable data input and (iii) it must be possible to comparethe output with a relative complexity scale.
7 A manufacturing enterprise system contains a large range of diverse elements:material, parts, humans, data, documents, machines, processes etc. and the number of possible relationshipsbetween elementsis a function of that range. Judged from the perspectiveof designingor re-designingthey are undoubtedly complex systems.Modelling using diagrams has been used to visualise systemsand to aid understandingby simplifying and describing system elementsand the relationshipsbetween elements. Models can be static or dynamic and be representedphysically, mathematically or graphically.
This research is concern'edwith static graphical models that are characterisedby the foflbwing:
* They are used as a meansof describingcomple)dty for investigationor problem solving.
4P They can be decomposedto show differqnt levels of detail.
* They can be usedto describea systemfrom'many perspectives.
4P They are produced with a definition of context that establishesthe boundary and scope of the systemunder investigation.
9 They Cannotbe all embracingand will have a viewpoint that wiH emphasisesome aspects of reality at the expenseof others.
In this researcha model is consideredto be a static graphical description using a formal documentedsyntax constructed from a givenviewpoint for a givenpurpose.
Various approacheshave been used to model manufacturing enterprise systems using diagrams,the impetusto. use modelsderives from a generalrecognition that they can provide relatively low cost, verifiable descriptionsof complex enterprisesystems to provide a basis for systemanalysis, improvement and design. A survey (Colquhoun et al., 1993) has shown that Information System(IS) software analysisand design methodologiesand Requirements Definition approachesare often the basis of modelling methods however the exhaustive
S detail required by IS approachesis often not necessaryto the same extent in enterprise modelling 'it is at the broader functional levels that there is the greatest potential for improvement' (Shunk et al., 1986). Modelling methods in manufacturingenterprises have received added impetus in the 1990's with the emergence of Business Process Re- engineering (BPR) as an acknowledged area of activity utilising modelling extensively (Jennison,1994).
A thread of enterprisemodelling evolution began in 1957 as part of the developmentof the Automatically ProgrammedTool (APT) symbolic programminglanguage (Kief and Walters, 1992) at the MassachusettsInstitute of Technology in 1957. Ross (1994) claims that the concept of graphical StructuredAnalysis (SA), the 'box notation', later to becomethe basis of the Structured Analysis and Design Technique(SADT) was first used in the project. The SA approach was subsequentlydeveloped by the US Air Force and Boeing to become SADT (Ross and Schoman,1977) and finally evolved into the IDEFo and IDEF, functional and data modelling methods (US Air Force, 1981a, b) in 1981. IDEF modelling methods have substantiallyinfluenced the researchpresented in this thesis.
1.4 IDEF, FUNCTIONAL MODELLING
IDEFOis a functional modelling method that uses graphical and text conventionsto describe functions or activities and their relationshipsin terms of information, objects, material and resources. An IDEFOmodel consistsof a seriesof diagramsand associatedtext describinga system. Diagrams are numberedto indicate their relative position in a structured hierarchy, upper levels of the hierarchy are general descriptions that become more detailed as the hierarchy expands(see figure 1.1).
The methoduses principles established for softwaredevelopment, those of Abstraction, Modularity, and Iliding. Abstractionallows the groupingof commonproperties of a system,Modularity (or decomposition)allows the division of componentparts of a systemand the definitionof relationsbetween parts and the conceptof Iliding allowsthe displayor examinationof onlythe levelof detailof interest.
9 In practice this means an IDEFO model can provide a means of representing the decomposition'of a system(consisting of men, machines,material, products etc.) into easily understood, related elementsusing a series of diagrams and text to the level of detail required by a user. Ross (1977a) advocatesthe use of IDEFo as a structured modelling method for manufacturingsystems in his proposal 'IIDEFogreatly increasesboth the quality and quantity of understandingthat can be effectively and precisely communicatedwell beyond the limitations inherentlyimposed by the imbeddednatural or formal languageused to addressthe chosensubject matter'.
Ifigh level diagram
I
Increasing detail
Figure 1.1 I-lierarchicaldecomposition - ICAM's architecturefor manufacturing (Source- Wisnosky, 1979)
1.5 RESEARCH OBJECTIVES
This researchis concernedwith enterprisemodelling as a meansof managingcomple)dty that provides a systemic approach to present a whole picture and one that is not specifically focused on information engineering.
10 The aim of this researchwork is to further the use of IIDEF modeffingmethods as a meansof gaining insight, designing or improving performance in manufacturing enterprise systems. The following objectivesare being pursuedto achievethat aim: e To advancethe understandingand use of modeRingmethods in the context of manufacturingenterprises.
9 To research potential areas for the improvement.and synthesis of IIDEF modelling methodswithin a manufacturingenterprise framework.
To test and validate improved modelling methods.
1.6 RESEARCH METHODOLOGY
Recognitionof the sipffiamee of modeffingmelhods
A surveyof publishedresearch on IDEFOand related modelling methods
A comparativereview A reviewof the significanceof An investigationof of modellingmethods systemarchitectures applieduse I*I Proposalsfor Proposalsfor a graphicalstructure aA surveyof IDEFO to describemodelling methods manufacturin enterprise practitioners framework
Proposalsfor a nationalsurvey Proposalsfor an 'Wz behaviouralmodelling methodI
Avvlicationof themethod A COATRL6U77ON TOJUVOR=GE ------A generic approach The manufacturing Refifiement and v A snapshotof the SERC funded to describing enterprise validation of Me use of IDEFO national survey modelling methods framework newmethod
Figure 1.2 The structure of the researchmethodology
11 The structure of the researchmethodology employedis shown in figure 1.2. The initial survey of published work identified the lack of a coherent, comprehensivebody of knowledge related to the use of IDEFo or other comparablemodelling methods used in manufacturingsystems in the UK and led to three avenuesof investigation:
(i) A comparativereview of the modelling methodsthat have been identified at the survey stage as those that have the potential to contribute to modelling in manufacturing enterprises. From this work came proposals to enhancethe scope of IIDEFofunctional modelling to include aspectsof systembehaviour.
(fi) The role of system architecturesand the need to develop an approach that clearly identifiesthe role of modellingmethods specifically in the context of manufacturing I enterprises. iii) An investigation of the use of IIDEFOand other modelling methods in manufacturing enterprises.
As a result of subsequentresearch a comparative analysis of modelling methods and a survey of the use of IIDEF methodsin the UK was carried out. A manufacturingenterprise framework was proposedand aspectsof two modelling methods,IIDEFo and IIDEF3,were integrated to develop a new modelling approach. The new approach was refined ml a manufacturingsituation and a methodology to support the new approachwas proposed. Finally a casestudy site was usedto refine and validate the proposals.
1.7 CONTRIBUTION TO KNOWLEDGE
The contribution to knowledgearising from the researchis in eight related areas: 1. A comparativeanalysis of modelling methodshas been provided. 2. An IIDEForeference model for manufacturinghas been proposed and validated 3. A survey of IDEFo practitioners has been carried out to provide a picture of the way the method is used in the UK.
12 4. The first publishedUK application of the use of the IDEF3 method in manufacturing hasbeen provided. 5. A manufacturingenterprise modelling framework hasbeen developed. 6. A meta-modelto describethe structure of modelling methodshas been developed. 7. A new compositebehavioural modelling method has been developed.
1.8 VALIDATION OF THE RESEARCH
As a meansof validating the researchIDEFO, IDEF3 and IDEFO-3 modelling has been carried out in six manufacturingcompanies: - Lucas Aerospace,Shaftsmoor Lane, Birmingham. Scholl ConsumerProduct, Derby. CannonIndustries, Wolverhampton. British SteelDistribution, Leeds British SteelAutomotive Centre,Wolverhampton ABB NOrnberg,Narnberg, Germany
Seventeenpapers (fisted in appendix 1) dealing with aspectsof this researchhave been produced, four havebeen publishedin internationaljournals, one in a UK journal, one in a Germanjournal, two at international conferences,six in UK conferencesand one in a book. One fin-ther international journal paper has been accepted by referees and is awaiting a publication date and one UK conferencepaper is awaiting refereesreports. In addition five paperson relatedtopics havebeen published during the research.
Throughout the work supporting studies were pursued in the areasrelevant to the i body of researchand ten conferencesin the areaof manufacturingsystems were attended over the period of the research.
In an initial phaseof the researchpublished literature on manufacturingmodelling, primarily IIDEForelated, was analysed,as was the use and application of modelling methodsin general. A key finding was the absenceof information relating to the way that modelling is used by
13 UK manufacturingenterprises to managechange. The evidencefor this finding was used to pursue a researchgrant to fund an investigation of. 'Best practice modelling methods in manufacturingenterprises". A grant of L120K was awardedby the Engineeringand Physical in Sciences Research Council - Control Design and Production group (EPSRC-CDP) October 1994 to identify and develop 'Best Practice' modelling methods in manufacturing (EPSRC-CDP grant ReferenceGR/K39400) and the research(not reported in this thesis) commencedin April 1995. The researchannouncement is shown in appendix5.
In 1993 the IIDEF3 method was recognised as having the potential to contribute to this research. As a result the author approachedthe US vendors of the IDEF3 tool to support the work and subsequentlyobtained a grant of $5840. The letter of agreementis shown in appendix6.
During the courseof the researchthe author has actedas a reviewerfor one book (Wu, 1994) and as referee for two papers in the area of systemsand IIDEF modeffing for the InternationalJournal of ComputerIntegrated Manufacturing.
S
14 CHAPTER 2
THE IDEF METHODS
2.1 INTRODUCTION
This chapter provides a historical perspective of the IDEF methods developed under the ICAM programme. Those IDEF methods pertinent to this researchare introduced and the latest generationof IDEF methods developedby Knowledge Based SystemsInc. are briefly reviewed.
To augmentthis chapter appendix3 provides a review of the syntax and semanticsof all the relevant IIDEF methodstog ether with a discussionof their role and a review of the software tools availableto support IDEF modelling.
2.2 A REVIEW OF THE ICAM PROGRAM31E
The United StatesAir Force initiated a master plan in the early 1970sas a long term strategy to control and co-ordinate aerospacemanufacturing. The work was carried out by the US Air Force Materials Laboratory as the Air Force Computer Aided Manufacturing (AFCAM) project (US Air Force, 1974). AFCAM was estabUshedas the first step in a strategy to
15 support military and space research manufacturing and technology growth into the next centuiy.
The subsequent development of the initiative was the Integrated Computer Aided Manufacturing (ICAM) programme (Le Clair, 1982). In 1978 the ICAM programme was formalised with the objective of realising the benefits of integrating the sub-systems.of manufacturingand extending the level of technology availableto aerospacemanufacturers. The perceived need nationally being to establishthe long term supremacyof the American aerospaceindustry and to transfer the knowledge gained to other domestic manufacturing industries.
Wisnosky (1979), manager of the ICAM project, describes the manufacturing systems integration conceptsadopted by the ICAM program as a structure of inter-communicating elements or functions in the form of a hierarchy, in ICAM terms 'An Architecture for Manufacturing' that he representedas figure 1.1.
In figure 1.1 each,truncated pyramid representsa level of detail on the top face and a decomposition.into smaller levels of detail on the bottom face. For instance the 'whole system' is composedof sub-functions1,2,3 and 4. Functions 1,2,3 and 4 have their own sub-functions,in this way the 'architecture' can extendto any level of detail. The hierarchy consists of 'Executive control' at the top levels followed by 'Management control', 'Technical support, 'Process control' and 'Direct manufacturing' at subsequentlevels. The diagram reproduced in figure 1.1 is now commonly used to represent the concept of hierarchical decomposition and many authors have used it to explain the principles of the IDEFomethod (Hughesand Maull, 1985a, Banedeeand Al-Malild, 1988,Wu, 1994).
An interesting developmentof this description of ICAM Architecture is the description used in a later US Air Force (1981 a) report 'ICAM Architecture'. Here the icAm Architecture is describedas a 'definitionof manufacturing'consisting of threemodels: A functionalmodel the 'blueprint', an informationmodel the 'dictionary' and a dynamicmodel the 'scenario.
16 The concepttakes account of different approachesto manufacturingin 'factory views' of the arcifitecturethat can representhow individual plants carry out manufacture. Together with a 'composite view' of the architecture to represent an 'industry wide aggregationof several factory views'. This concept is used consistently in post-1980 ICAM publications for instancein the IDEFo Architects Manual (Ross et al., 1980) ICAM architecture is defined from a functional perspectiveas: 'blueprints, drawings and specificationsthat constitute a . formal definition of manufacturing-a model which is a logically consistentand demonstrably accurate representation of the functions of manufacturing and the relationships between those functions' The composite view is describedas representing,current practice for the production of a single, new, major aerospaceproduct, such as an aeroplane,' 'emphasising the essentialfunction, information and material flow necessary'.
The work carried out in the ICAM project was sponsoredby the US Air Force primarily for the aerospaceindustry Jones (1979),. an ICAM program manager,explained: 'the ICAM program neededa factory that was like most targeted aerospacefactories for setting scope and for integrating ICAM sub-systems'.Funding preventedthe physicalbuilding of a factory so an architecture of manufacturing using models of specific factories was produced and those factory views were then merged into a single compositeview of manufacturing. The composite model was reviewed by an industrial panel with the result that 'it gave us confidencethat the architecture of manufacturingrepresents manufacturing in general and not aerospacemanufacturing only' (Jones, 1979). The model was releasedinto the public domain in 1980 (CAM-I).
An essentialelement of the ICAM approachwas the drive to 'establishstructured methods and tools for applyingcomputer technology to manufacturingand to demonstrateICAM technologyfor transitionto industry'(US Air Force,1981a). In orderto meetthe objectives of identifyingcommon functions, information requirements and usageacross the aerospace communitythree modelling approaches were identified. After 'a reviewof over fifty systems modellingtechniques' (Mayer, 1979) the Functionalmodel, the Informationmodel and the User Interface(dynamics) model were consideredessential to supportthe decisionmalcing
17 processrequired for integrated system development. The Structured Analysis and Design Technique(SADT) was selectedfor functional modelling as a first step.
2.3 A HISTORICAL PERSPECTIVE OF THE IDEF METHODS
2.3.1 A review of the development of IIDEFo
The concepts underlying SADT stem from Ross (1977a, b) and his Structured Analysis thinking in the development of Automatically Programmed Tools (APT) machine tool programming language and Hori's (1972) 'Human-directed activity cell model' with philosophicalfoundations in Ackoff s 'Scientific Method' (1962).
The 'box notation', later to becomethe basisof SADT, was first used by Ross in developing APT at the MassachusettsInstitute of Technology in the late 1950s. The concept of 'hierarchical decomposition' was also recognisedby Ross in the early 1960s but the two fundamentalelements were not formally linked until the emergenceof Ross's 'structured analysis' in 1969.. Ross's seminalpapers (Ross, 1977a,Ross, 1977b, Ross and Schoman, 1977) describethe foundation of Structured Analysis and SADT thinking.
SADT was developedjointly by BoeingInc. andthe companyRoss and Schomanfounded, SofrechInc., andwas funded by the AFCAM project. In an accountof the developmentof SADT Ross(1985) explains that: 'SADT brokenew ground in the areasof problemanalysis, requirementsdefinition and functional specification because it allowedrigorous expression of highlevel ideas that hadpreviously seemed too nebulousto treattechnically'.
The technique SADTTM was commercially availablefrom Sofrech by 1975, early reports of its use and potential are given by Combelic (1978) and Ross and Schoman(1979). In the
is first phaseof the ICAM project from 1976n8 the US Air Force continuedto use SADTTM under licence however as Brovoco (1991) explains 'ICAM part 2 came up for bid and the Air Force did not want to continue buying SADTTM licences. In addition they wanted to improvethe method.
Soffech Inc. agreedto develop a public domainversion of SADT incorporating the changes required by the ICAM project (the inclusion of function lists (node index), Information matrices (Data dictionary), AS IS and TO BE concepts, Factory Views, Composite and Genericmodels and the deletion of data modelling) in order to win the ICAM part 2 contract and in October 1978 Ross, Feldmanand Mayer beganthe revision of SADT (Mayer, 1993a). Wisnosky and Ross used the acronym IDEF (The I-CAM DEFinition technique) for the Wisnosky (1979) 'With it (ICAM 2) has revised method, explains part - also come an integrated system defining both vocabulary and language,called IDEF, to facilitate inter- communication and the description of manufacturing operations. The resultant common terminology will allow future software deliverablesto be mutually compatible and easily integratedinto ICAM-establishedtechnology'.
The new method was termed the IDEFo Function model to differentiateit from other aspects of the ICAM part 2 project that included the developmentof IDEF, an Information model and IDEF2 a Dynamics model. In parallelvvith these developmentsBoeing Inc. were given the task of designingAUTOIDEFo, a software tool for IDEFo.
The IDEFo method was approved for public release (in the USA) in February 1980 by Wisnosky the ICAM project leader. The IDEFo 'Architects Manual' (Soffech Inc., 1980) was publishedin February 1980. The more widely used manualfor 11DEFowas releasedinto the public domain by the US Air Force through the Wright Aeronautical Laboratories in 1981 (US Air Force, 1981a). The IDEFo method is now supportedby a range of software tools (discussedin detail in appendix 3 section A3.3) and has an establishedrecord of successfuluse particularly in the. USA. In May 1992 The Institute of Electrical and Electronics Engineers (IEEE) accredited IIDEFo and releasedan IEEE/IDEFo standard in
19 February 1995. In 1992 the National Institute of Standards and Technology (NIST) authorised a programme to establishIIDEFo as a Federal Information Processing Standard (FIPS) for formal releasein April 1995. Sudnick (1993) notes that an objective is to obtain intemational (ISO) acceptance.
In the UY, the first recorded use of IDEFOwas by Crossley,Davies, and Richardson (1978, 1979) in ProgressReports to the Air Force Office of Scientific Researchin the USA. These two reports formed the first two deliverablesof an ICAM contract the 'Preliminary Design of a Job Shop Control System' that was undertakenby the University of Salford Industrial Centre. It was 1984 before referencesto the method began to appear in UK research publications and a survey carried out by Colquhoun and Baines (1993a) in 1992/3 revealed over fifty UK and Europeanpublications relating to IDEFo.
2.3.2 The development of associatedIDEF methods
The IDEF methods 0,1 and 2 were envisaged as an integrated 'Architecture' for manufacturing.and Soffech were given the responsibility for the 'Integration of ICAM modulesand the orderly transition of ICAM modulesinto ICAM systems'in effect they were responsiblefor documentationof the ICAM modellingmethods. Their final report on ICAM architecturewas publishedby the US Air Force in 1981 (US Air Force, 1981 d).
ICAM's information model, IDEFI, was developedby the Hughes Aircraft Corporation (US Air Force, 1981b) and IDEF2, the dynamics model, was developed by Higher Order Software (HOS) Inc. and finally by Prisker and Associates(US Air Force, 1981c). Both were approved for publication in 1981 along with IDEFo however they were subject to For Early Domestic Dissemination(FEDD) control and export of the documents-wasconsidered a violation of US Traffic-in-Arms regulations. The control was in force until 1982 for IDEFo and 1983 for IDEF, and 2. In discussingthe diversity of sub-contractorsto the ICAM
20 project Brovoco (1991) notes 'Is it any wonder why IDEFO,I and 2 are not totally integrated. In fact the IDEF, and IIDEF2developers were never required to be BDEFoexperts'.
In 1983 an Integrated Information Support System (IISS) project was initiated under the ICAM program focused on the 'capture, managementand use of a single semanticdefinition of the data resource referred to as a Conceptual Schema'. The project carried out by D. Appleton Company (DACOM) developed and extended the Logical database Design Technique (LDDT) and the IDEF, method and releasedit as IDEFIeXtended in 1986 (US Air Force, 1986).
A review of commercial.modelling tools supporting the IDEF methodshas been carried out as a precursor to the survey detailed in section 3.5 and is presentedin appendix 3 section A3.3. It is not the intention, in this research,to comparethe use of the various tools but to establish the extent of computer support for IIDEF methodsas part of the picture of the use of IDEFo and its relationshipwith other modelling methodsdeveloped in depth in chapter4.
In the USA a focal point for the discussionand developmentof IDEF methods has been the 11DEFUsers Group (now re-namedthe Society for Enterprise Engineering (SEE), (Preston, 1995). Research connected with 11DEFmethods seems largely concentrated on those universities with funding from US defence industry contracts notably the Texas A&M University. Interest in the 11DEFmethods generally appearsto be industry driven and is in direct contrast to the UK where the interest appearsto be academicdriven. A review of eleven USA IDEF User Group conferences from 1991 to 1995 revealed that eight universities presented papers and (in late 1995) the last three conferenceshad only six university delegates. All other attendeesbeing industry or governmentbased.
IDEF methodsand new IDEF initiatives are now under the control of the Logistics Research Division of the Air Force Armstrong Laboratory through their Information Integration for Concurrent Engineering (IICE) programme. In May 1992 the programme released an interim report (Mayer et al., 1992a) describing IDEF3, a Process Description capture
21 method. The method was developedin conjunction with Knowledge Based SystemsInc. (KBSI) who are the Air Force contractor responsiblefor the continued development and maintenanceof IDEF systemsand software. In the KBSI IDEF methods the identifying number (3,4,5 etc.) is used in standard script to distinguish it from the ICAM derived methodswhich use subscriptnumbers.
YJ3SIdeclare that 'the IDEF acronymhas beenre-cast as the namereferring to an integrated family of integration DEFinition methods' (Mayer et al., 1992a). Their involvement has led to significant developmentsextending the range of IDEF methods to encompassobject oriented design (IDEF4), ontology description capture methods (IDEF5) and design rationale modelling (IDEF6).
2.4 THE ]LATEST GENERATION OF IDEF METHODS
This section briefly introducesthe range of IDEF methodscurrently under development by KBSI who now the US Air Force contractor responsible for the continued -are developmentand maintenanceof IDEF methods.
YheIDEFS Ontology description capture method
Ontology is the study of what there is, describedby Menzel et al. (1992) as an 'activity at work acrossthe fidl range of human inquiry prompted by humanity's persistent effort to understandthe world in which it exists and which it helpedto shape'.He uses an example of ontology in the world of sub-atomicphysics 'to develop a taxonomy of the most basic kinds of objects that exist in the natural world; electrons, protons, muons and their fellows'. Perakath (1995) describes an ontology as 'a precise definition of the terminology in a domain'.
22 IDEF5 is an ontology description capture method with theoretical foundations in set- theoretic semanticsand first-order modal logic (Menzel et al., 1992). It is designedto capture and describe the basic entities that populate a manufacturing system such as personnel,machines, material etc.. The method involves; 9 Providing an inventory of the Ends of objectsthat exist in a domain according to the best sources of information regarding the domain, such as a domain expert or estabfishedtheory. * For each kind of object, providing a description of the properties common to all and only instancesof that kind * providing an inventory of the associationsthat exist within a given domain between (and within) kinds of objects.
The method is seenas a meansof providing a structuredlanguage that can be interpreted unambiguously,that can remove ambiguities in terminology and that will ultimately be amenable to automated reasoning. No formal links with other IDEF methods are proposedbut the method hasthe potential to provide a common pool of information from which other methodscan extract.
YheJDE-F6 Design rationale capture method
The 11DEF6method is a meansto acquire and representthe rationale behind the design of Information Systems to answer the question why? a particular decision, strategy or feature was adopted or an assumptionmade'. The method is in the concept stage and is seen as a 'series of adjuncts to other modelling tools rather than being a stand alone method' (Mayer et al., 1992c).
Future IDEF methods.
The IDEF methodsunder considerationin 1995 (Perakath, 1995) are: IDEF8 User interface modefling
23 IDEF9 Scenario-drivenIS design
IDEF10 Implementationarchitecture modelling IDEF12 Organisationmodelling 11DEF13Three-schema mapping design 11DEF14 Network design.
24 CHAPTER 3
PUBLISHED IDEFo AND IDEF3 RESEARCH
3.1 DaRODUCTION
This chapter reports the findings from researchinto published literature relating to IDEFo and IDEF3. The research followed four specific lines of inquiry: (i) To expose how researchersperceive the use and application of the methodsin the context of manufacturing systems.(ii) To establishhow use of the methods has been evaluated.(iii) To investigate proposalsto modify the methods.(iv) To analysehow the methodshave been used in a wide range of applications. The rationale was to provide a greater understandingof the use of modelling methods in the context of manufacturing enterprises, to investigate potential improvementsto existing methods and to investigate the potential to develop the use of modelling for manufacturingapplications.
Four broad categoriesof publishedresearch provide a frameworkfor examinationof the methods,they are: (i) Descriptionsand reviews of the basic principlesof IDEFo. (ii) Performanceevaluations of IDEFo and comparisonswith other methods.(iii) Proposed enhancementsof the IDEFomethod. (iv) Applicationsof the IDEFOmethod.
25 The researchestablished that, in contrast to the IIDEFOmethod, there is a dearth of published information related to the 11DEF3method, as a result researchfindings are presentedas a generalreview in section 3.4.
An initial literaturereview was carriedout in the first stageof the researchand published in 1993(Colquhoun et al., 1993a).Kusiak et al. (1994)acknowledged that the reviewrevealed gseveralissues that had receivedlittle attention in the literature'. It focused on the applicationof IDEFoin manufacturingsystems and gave 11itomi (1994), in citing the review, an opportunityto reinforcehis definitionof manufacturingsystems engineering as 'integrated manufacturingunifying material flow (manufacturingprocesses) and Information flow (productionmanagement)'. A key finding from this stageof the researchwas that there werefew publishedindustrial applications of the useof the 11DEFomethod. To identify and accessindustrial users of the methodthe leadingUK IDEFOsoftware Tool vendor' was approachedto supportan investigationof the useand application of the Design/IDEFTool, the findingsare detailed in section3.5.
3.2 STANDARD REFERENCE SOURCES
Section2.3 discussedthe role of SADT as the precursorof IDEFo and Marca and McGowan (1987) have provided a key publication for the study of SADT activity modelling. It contains a complete account of the concepts,method and author/readercycle methodology together with applications. Interestingly they do not include the data modelling aspectsof SADT which by the mid-80's had not been widely adopted and had been overshadowedby other data modelling methods based on DeMarco's (1979) Data Flow Diagram (DFD) approach. Little has been published on SADT data modelling however Ross (1985) provides an account of the basics and in 1994 he still advocatesthe advantagesof SADT 'Dual' activity and data modelling functionality (Ross, 1994).
1EDEFine Ltd., Crowthome, LTK.(Formerly NEcroMatch Ltd. ).
26 Three basic sources of reference are available for the study of IDEFo; the US Air force IDEFo report (US Air Force, 1981a), the Architects Manual ICAM Definition Method - IDEFo (Ross et al., 1980) and the recently published IDEFo Function Modelling (Mayer, 1993a). They have a similar format and provide a detailed description of the method; the rules, syntax, diagram and model format, text presentation as well as model validation, document control procedures,interview techniquesand guides to understandingand using activity diagrams.
Mayer's approachin 'IIDEFoFunction Modelling', although rigorous in describingmodelling syntax and semantics,is less formal in terms of describingthe author reader/cyclepreferring to call it teamwork discipline. 11is 'Guide to creating IDEFo diagrams' also provides a 0 practical guide to good practicein diagrammingmethods. The applicationoriented approach used throughout the book suggeststhat use of the method has been refined from experience since the previous manualwas produced in 1981. Hill (1995) has publisheda guide to the application of IDEFo which provides a detailed review of the method concentrating on practical applicationrather than methodology.
3.3 A STATE OF THE ART REVIEW OF IIDEFo
This ongoing literature review that followed the initial literature review in 1993 (Colquhoun, et al., 1993a) has re-enforcedthe finding that publishedwork relating to IDEFo continuesto fall into oneof four broadcategories albeit with differentemphasis:
L Papersthat deal with descriptionsand reviews of the basic principlesof the method. 2. Performanceevaluations, or comparisonswith other methods. 3. Proposalsfor enhancementsof the method. 4. Specific applicationsof the method.
27 3.3.1 Category 1: Descriptions and reviews of basic principles
Several manufacturing systemstextbooks now provide outline descriptions of the IDDEFo method. After describingthe principles of the method Wu (1992), for example,uses the method as meansof presentinghis ideas on material and information flow in manufacturing systems. Similarly Ranky (1992) describesthe method and usesIIDEFO diagrams extensively to presentideas. Bauer et al., (1994) briefly introduce SADT and use a structured set of diagrams to describe shop floor material and information flow. Waldner (1992) also introduces SADT as a methodology for CIM implementation. Mitchell (1991) introduces IDEFo and uses IIDEFodiagrams to describe a 'System-EnvironmentSimulation' approach. In contrast Rembold et al. (1993) use IIDEFodiagrams to describemanufacturing operations without explainingthe principlesof the method.
Mackulak, (1984) provides a detailed review of I-CAM and the role of 11DEFoand goes on to explain the principles of the method. His description includes an example of a node glossary, an aspect of modelling often overlooked in simple reviews of the method. Harrington (1985) was an early advocate of IDEFo who proposed it to help understandthe complexity of manufacturingprocess. He describesthe principles of the method and usesthe analogyof IDEFo providing a two dimensionalmap of manufacturingthat allows the human mind to deal with specific elementswhilst retaining overall relationships. His model is based on an environment diagram (A-1) 'Conduct a manufacturing enterprise' with examplesof decomposition down to the level of an A36 diagram. He also suggests two other applications for the method, the first is the use of a generic model of manufacturing to evaluate or establish deficienciesin an organisation and the second is the use of IDEFo models as a meansof planning the integration of manufacturing functions. He concludes with the claim that 'One of ICAM's greatest contributions to the scienceof manufacturing will prove to be the developmentand use of the structural [sic] analysismethodology called IDEF. The complete 'Conduct a manufacturingenterprise' model is presentedin his book on the subject of understandingthe manufacturingprocess (Harrington, 1984) the model has been subject of much interest by other authors Graves et al. (1989), for example, use Harrington's diagramsas a context for their proposalsfor a methodologyfor CAPP. Early
28 interest in IDEFo in the UK is characterisedby Hughes and Maufl (1985b) who proposed IDEFo as a meansof designingCIM systemarchitectures.
Other papersthat provide brief descriptionsof the principlesof the method include Goldman and CuRinane(1987), Yeomans (1987a), Parnaby (1988), Baines and Colquhoun (1990), Sarkis and Lin (1994) and Deng et al. (1990).
3.3.2 Category 2: Performance evaluations and comparisons
The bulk of published work is in this category reflecting the early investigative interest generated by IDEFO after its release into the public domain. Comparisons of various 'competing' methodshave beenused to investigatetheir role in systemsdevelopment andthe extent of their capability.
Buchel et al. (1984) carried out a detailedcomparison of 'Known methodsused in the design of production managementsystems' comparing four structured methods, IDEFo, SSADK S.E. (Structured Engineering ), and GRAL The criteria used for comparison being their ability to handle complex systems,their ability to describea system,formal aspectsof the methods and computerisationof the method. In concluding,the authors describethe GRAI model as a methodology for structuring a specific application field (production management systems)and strongly advocate its use proposing that GRAI 'represents an integration of three different viewpoints' 'the Physical system and its associated material flow, a hierarchicalmodel of the Decision systemand the Information system'. IDEFOis seenby the authors as 'almost exclusivelyconcerned with functional requirements'.
Industrial applications of the GRAI method have not been widely reported and published examplesof its use are largely in French companiesfor example;SNECMA, SNPE Toulouse and SMG Cededur Pechiney (AUGRAI, 1987). In the Ul-ý Ho and Ridgeway (1994)
29 describean application in the design of cellular manufacturingand Ridgeway and Downey (1991) and Ridgeway (1992) haveproposed other applicationsfor the GRAI methodology.
Maji and Stevenson(1988) discussformal approachesto the developmentof information system specifications. The paper focuses on a comparison of IDEFo and the Structured SystemAnalysis Design Methodology (SSADM) (Longworth and Nricholls1987), the author proposes that IDEFo and the Data Flow Diagram (DFD) of SSADM have similar basic principlesbut that in a DFD the source and destinationof data is shown,whereas in IDEFo it is difficult to understandthis aspectof the model. The author's criticism can be countered by the use of a node index (Colquhoun et al., 1989) however. The author seesIDEFo and SSADM as two of many approachesthat include the GRAI method (Pun et al., 1985, Doumeingts.,1985), the Soft SystemsMethodology (Checkland, 1981) and SADT. He concludesthat the methodologiesmay be suitablefor the developmentof information system specificationbut that further researchwork is requiredin various areasof the methodologies. A brief comparisonof the same methods, GRAI, SSADM and IDEFo is also provided by Wyatt and Al-Malild (1990).
Roboam et al. (1989) however advocate the need for an integrated methodology for the analysis of production systems claiming that no single methodology exists for physical, decisional and informational aspects of systems. The authors select IDEFo as a complementarymethod to GRAI (Doumeingts, 1984) and Merise (Rochfeld and Tardieu, 1983). They propose that IDEFOcan provide a functional model, defining the elementsof the physical system and the flows of information and objects between them. Whereasthe GRAI and Merise methodsare capableof analysingand designingthe managementdecision system. Roboam and Pun (1989) have surveyedand classifiedmanufacturing system design methodologies. Their work draws no definite conclusionsregarding the role of IIDEFobut re-enforcesthe view that it is not possibleto use the methodologyfor the analysisand design of both the physical and decisionalsub-systems. Wood and Johnson(1989), compareIDEFo and SSADM and also conclude that no single method offers the complete answer for manufacturingsystems analysis.
30 Using an experimentalmethod basedon users, Yadav et al. (1988) carried out a comparison of DIFDs and IDEFo in terms of requirementsdefinition. Four criteria for comparisonare used: syntactic, semantic,communicating ability and usability. The authors report that the groups of analystsinvolved in the study had confidencelevels (in their own EDEFomodels) of 82.5 % in syntactic correctnessand completenessand 72.5% confidencein their semantic accuracyand analystsusing IDEFo had higher confidencelevels in their models than those using DFDs. The authorspropose that the work is inconclusiveand that it failed to establish which method produces a better result. They conclude that, although the DFD method is easierto learn and use, further work is required to establishwhich method producesa better model.
An assessmentof all the IIDEF methods was carried out by Godwin et al. (1989) using a methodologydesigned to assessBDEFo as a 'descriptive tool'. The work concludesthat : 'It is certainly one of the main strengthsof the method that, althoughpossible, it is very difficult to produce an IIDEFOmodel which is correct (in EDEFoterms) but inaccurate as a system description'. A complementarystudy by the sameauthors (Godwin et al., 1988) evaluates the IIDEFoas a 'sub-model' of a complete IIDEF model consistingof functional, information and dynamic sub- models. This is one of the few publications that provides a detailed discussionand an example of the integration of the first three IIDEF methods, reviewing consistency across the sub-models, and assessingthe strengths and weaknessesof a combinedmodel.
Ming Wang and Smith (1988) contrast the Soft SystemsMethodology with computer based IIDEFoanalysis and propose that the two approachescompliment each other. The authors emphasisethe ill-structured 'soft' nature of manufacturingand the needfor those developing information systemsto start any analysisat a conceptuallevel. The ability of computer based IDEFo to enhancethe quality and consistencyof diagramsand models and improve system modelling productivity is recognised. In contrastingthe two methodsthe authors seeIDEFo as techniqueorientated in contrast to the 'sophisticatedand mature' Soft Systemsapproach which requires experimental analysis and high intellectual input. The authors advocate
31 incorporation of the Soft Systemsapproach into emistingIDEFO software to combinethe two approaches.
Maull (1986) evaluated the contribution IDEFo can make to the analysis and design of Computer Integrated Manufacturing Systems. I-Es review examines application of the method in a number of manufacturing situations and concludes that his research demonstratesthe usefulnessof the method as a 'simple and effective communicationtool'. IEs research also highlights some criticisms of IDEFo in its ability to describe certain characteristicsof systemssuch as time, dependency,data-entity relationshipsand reliability requirements. In addition, along with other authors,he advocatessoftware tools to simplify the model building process.
Other relevant evaluationsand comparisonsof the method have been carried out by Rzevski and Maji (1988) and Maji and Stevenson(1988).
3.3.3 Category 3: Proposed enhancementsof the method
Much publishedresearch has been focused on modifyingthe methodto overcomeperceived f0ings, to extend its capabilityor to integrateit into methodologiesthat utilise other modellingmethods. ý This aspectof IIDEForesearch was evidentbefore the methodwas matureand beforeIIDEFo computer tools were.widely available. Over the last five years howeverthe numberof publicationsconcerned with enhancingthe methodhas increased. This sectionhas been split into two areas;proposals to modify the IODEFomethod and proposalsto combineIODEFo into othermethodologies.
(i) Proposalsto modify the IDEFo method
Shunk et al. (1986) proposed a modification to the basic model format to overcome what they saw as a major drawback of the IIDEFo model in that. the large amount of data
32 generateddefies the sorting and assimilationefforts of many systemdevelopers, the user may not, therefore seea particular processas part of a larger CIM system'. As an alternativethey propose a Triple Diagonal technique(IDEFO-TD) basedon information, control and material flow that, it is claimed, provides a bottom up facility for model development. The method also identifies feedback information by modifying the diagram format to use colours and different line-types*todistinguish classes of ICOM.
Mandel (1990) describesthe methodas a meansof 'graphicalprocess description' using a modelof printedcircuit boardmanufacture. In his descriptionof the capabilitiesof IDEFohe recognisesthat it is capableof representinga processactivity and the informationand materialsinvolved in the process(what is happening)and proposesthat the methodis not focusedon organisationalstructure (who is doingit), howeverhe later clarifies the role of the 'mechanism'as an indicationof 'who' is carryingout the process. Threeshortcomings of the methodare cited: The amountof informationcontained in a diagramcan cause clutter,that the methodologylacks the ability to relatean interfaceitem with its originating activity directly andthat the methodologylacks clarity in defininghierarchical and interface * relationships.He proposes'ports' with text descriptionsto identify sourcesand destinations of informationcrossing the boundaryof a diagramtogether with a modified format to enhancediagram clarity.
taba (1992,1994) proposesthe use of different TOM arrow line-types to differentiate between passive physical objects, command information and data. Other authors have proposed changesto ICOM arrows Marran et al. (1989) for exampleuse four different line- types to show material flow, commandflow, resourcesand information flow.
Feller and Rucker (1990) propose extensionsto IDEFo to 'allow additional classes of knowledge useful to a communicationanalysis to be represented'. The authors propose a modified (eight sided) activity box to explicitly identify feedback information, controls and triggers in order to MCOrporatethe concept of processactivation or firing' using Petri-Net principles. Niel et al. (1992) propose an extension to SADT in a Failure Mode Effect
33 Analysis (FMEA) (O'Connor, 1991) application using 'Entries' to represent Inputs and Controls and using activities to indicate a sequenceof events. Kusiak et al. (1994) propose an approachthat 'violates the basic premise of IIDEFomodelling' using a model containing 27 activities and representingthem on a singlelevel diagramas a sequencedprocess graph.
Ang and Gay (1993) use IIDEFo as the basis of a modelling method for project risk assessment-theypropose modifications to model syntax, annotations to diagrams and diagramlayout changesto describesystem characteristics such as the probability of activities occurring, the duplication of tasks, and activity activation.
The use of computer supported IDEFo diagram production was considered an essential enhancementto the method before such tools were widely available. The principles behind the first IDEFo software tool developedby Boeing for ICAM part 2 are describedby Smith et al. (1980). Hartrum et al. (1988) describes'SA tool' a Sun workstation based graphic editor system with the enhancementof a data dictionary. Gamble (1988) developed a prototype IDEF o tool interestinglyusing IIDEFoto model his software before coding.
(ii) Proposalsto combineIDEFO with other methods
Bachert et al. (1983) proposedincorporating IIDEFOwith the 'System Analysis of Integrated Networks of Tasks' (SAINT) simulationsystem., In their proposal IDEFo was used to model 'Tasks' as IDEFOfunctions eachTask then becamethe subject of a dynamic simulation. An interestingvariation on the use of 11DEFmechanisms was their concept of using algorithms as mechanisms.Research finking 11DEFoand SAINT is also describedby Stockenberg(1986).
Wang and Fulton (1994) used IDEFo as part of an object-orientedapproach to information systemdesign. The approachincorporates DFDs and Entity relationship models where 'the DFDs and IDEFOdiagrams are comparedfrequently to find any discrepanciesin the models". Snyder et al. (199 1) combineIDEFo with 'Concept maps' and 'Storyboards' to overcome a weakness the authors claim IDEFo has 'for modelling extremely complex and dynamic
34 enviromnents where boundaries are fluid and the task priorities, conditions and interrelationshipsare continuouslychanging'.
For ESPRIT project 2439 (ROCOCO) Los et al. (1992) used IDEFo in conjunction with NLAM (Nijssen and Halpin, 1989) data modelling and the GRAI method (Doumeingts, 1995) in order"to model 'operation, data and control' in a single methodology. In their methodology an IDEFo model was linked to a higher level GRAI model using the data flows is in (1993) of a NLAM model. -Their approach contrast to Zgorzelski and Zgorzelska who also advocate the integration of NLAM and I1DEF0their 'NIDEF' methodology which adds graphical symbols from NLAM to 11DEFodiagrams. Their justification for this approach is not to improve 11DEFobut to replace NIAMs functional modelling feature which they consider 'significantly weaker than SADT or 11DEFo'.
The 'IFEM' methodology proposed by Malhotra and Jayaraman(1992) is one of the few publications that proposes integration of IDEF methods. In their approach ICOMs in an EDEFomodel becomeentities in an IDEFLx model and the IDEFo model is used as a dynamic model by 'extending it to model the temporal interactionsbetween activities'. An example of a 'dynamics description script' for an activity is usedto demonstratethe extensionfrom a functional model to a dynamicmodel.
In an exposition of ESPRIT project 688 Jorysz and Vernadat (1990a) give an account of the role of IDEFo principles in the requirements definition phase of the CIM Open System architecture (CIM-OSA). The authors maintain that CIM-OSA 'goes far beyond previous modelling tools and CIM system methodologies' and claim that: 'CIM-OSA produces a processablemodel of the CIM system as opposed to SADT-based methods or to IIDEFo which only produce static or incomplete descriptivemodels of system requirementslacking dynamicsmodelling, preciseinformation modelling and technical implementationissues'.
Other methodologies have been proposed using 11DEFoas a functional modelling element include: LUM-TM (Kehoe et al., 1991), II)EM (Wang et al., 1993), SIMCON/2 (Siggaard
35 and Bflberg, 1992), IMP (O'SuHivanand Browne, 1993), 'The Five Step Method' (Pageand Saunders,1993). MOSES (Molina et al., 1995).
3.3.4 Category 4: Applications of the method
Papers in this category demonstratethe diversity of applications in manufacturing. Ross (1985) in a general review of the use of IDEFo claims that 'Thousands of people from hundreds of organisaflons working on more than one hundred major projects use the methodology not only for the technicalwork'of systemdefinition and design,but for project managementand integration as well. The ICAM Program Office has cited IIDEF as recipient 0 of a Top-of-the-Line Manufacturing Technology Success Story award. The IDEF methodologiesare taught in severaluniversities. '
Komanduri et a!. (1985) selectedIDEFo as a suitablemethodology for determining machine tool systemrequirements in a high speed/high-throughputmachining application. The work uses a viewpoint diagram AO, node tree and provides a discussionof the use of the model down to level A22433. Bowden and Browne (1991) use a model based on an Al diagram 'Design the Production Environment" to examineinformation integration in the design of manufacturingfacilities. The model is decomposedto levels Al 13, A123 and A133 and the information flow described by the diagrams forms the basis of the authors analysis of activities involved.
Several applications of IDEFo in the field of quality assuranceare available..Stephans and Fox (1987) presenteda method for assessingquality systemsthat, the authors claim, 'should provide a' complete picture of requirements resources procedures, people, actions and interactions resulting from the IDEFo technique forcing'a systematicthorough evaluation. Tannock and Maull (1988) proposea strategyfor the developmentof quality systemsin CIM using IIDEFoand Maylor and Butler (1991) propose IDEFo as a suitable tool to model an existing quality system and as a means of obtaining maximum benefit from new system
36 implementation. Barton (1994) provides a design for a reliability testing system for microwave devicesusing an 'IDEFo representationof the testing process,which is easierfor the database novice to understand and implement than the usual data flow or entity- relationshipdiagrams.
Harrison et al. (1988) proposeIDEFo as a way of establishingthe functional specificationfor CIM sub systems,the work is basedon an applicationat NCR (Manufacturing) Ltd Dundee. The activity diagrams are decomposedto level A42 describing the PCB manufacturing system. They see the method providing an insight into system behaviour and forming a functional CIM integration specification. In a similar application Banedee and Al-Maliki (1988) propose a method for applying the IDEF methodsto design a flexible manufacturing system. They reinforce the view of others that no single method can provide a meansof developingsystem specifications but by exploiting the merits of severalmethods an enhanced modelling methodology can be developed. The applicationis illustrated with relevant IDEFo diagramsand the links betweenfunctional and information modelsis explained. Pandyaet al. (1990) also use IDEFo as a method to model the functional aspects.of a flexible manufacturingsystem and identify information which is then used to build an IDEFIx model of information requirementsat cell level. Using IDEFo models in CIM applications to describeinformation flow and subsequentlybuild data modelsis also advocatedby Chadhaet al. (1990,1994) who use the IDEFOand DFD modelsto provide a functional description of mechanicalhandling and ExtendedEntity Relationship(EER) diagramsto build a data model basedon the information definedin IDEFo diagrams.
3.4 A REVIEW OF PUBIJSHMD IDEF3 RESEARCH
The key referencework for IDEF3 is the D)EF3 Processdescription capture method report (Mayer et al., 1992a), a commercialversion of the report is also available (Mayer, 1993b). The report is a description of the semantics,syntax, scenarioconcepts and the methodology associated with the method. Unlike IIDEFo, public release of the IDEF3 method was
37 followed immediately by a software tOO12and a review of the method is provided in the software systemusers guide (KBSI, 1994).
In an early account of IDEF3 by Mayer (1990a) the semanticsof the method are explained and a prototype of the IDEF3 software tool (later to becomeProSim, marketedby KBSI) is described. In reviewing use of the method Huff et al. (1991b) claim that IDEF3 process descriptionsare simple enoughto be read and maintainedby 'everyone in the organisation'. The thrust of their researchis the dynamicrepresentation of EDEF3descriptions using Petri net models. The authors provide examplesof the conversion of an Object State Transition Network (OSTN) and a ProcessFlow Network (PFN) to a Petri net. For examplethe logic describedby a PFN asynchronous'AND' junction before a processis converted semantically into a Petri net 'Place' (representinga situation that, if conditions are met, the process is ready to take place). It is proposed that using the approach 'the Petri net models can be used to verify the performanceof a process. Cullinane and Mayer (1992) propose the use of 10DEFoand II)EF3 to improve the control of warehouseoperations. IIDEFois used to identify activities involved in operation of the system activity descriptions which are then combined with control charts to track performance. IDEF3 diagramsare used to establish methodsfor bringing the systemunder control, the authors claim that 'IDEF3 has proven to be an excellent method for describing an ordered sequenceof activities as they should be performed' and that an analystcan 'walk through the 11DEF3description seekinga reasonfor out of control points, offering explanationsand proposing potential solutions'. The authors treat the diagramsas separateentities and offer no formal syntacticallinks between II)EFo, EDEF3 and control charts. Benjamin et al. (1992) propose a knowledge based simulation architecture in which an 11DEF3Tool provides the systemdescription aspect of simulation. Their architecture is the concept behind 'KBSI's evolving Integration Platform' that is designed to provide an operating environment allowing different tools and systems to communicateexchange data and sharefunctionality. The examplegiven by the authors is of a simulation that could extract current schedulingdata from a factory information systemto use as input to a simulation with an IDEF3 model and elaboration documentsto hold the systemdescription.
2ProSim. and ProCap, Automated ProcessModelling, (KBSI, College Station, Texas).
38 One aspect of KBSI's researchaimed at interfacing IDEF3 and simulation systemsis the ProSinvVITNESS simulation link currently marketed in the UK by AT&T ISTEIý and described in 'Generate your WITNESS simulation models automatically with ProSim' (AT&T ISTEL, 1995).
Cesaroneand Dobrzeniecki (1992) describe researchfunded by the US Defence logistics agencywhere IDEF3 models have been used to investigatethe 'complex value-addedchain of sub-contractors'used in the manufactureof military aircraft by Lockheed. The authors describethe IDEF3 method as a 'transition network' that can be used to provide critical path analysisand activity basedcosting. Their approachinvolves the use of IDEFo, IDEFix and IDEF3 as discrete methods however the potential for integration is not discussed. Benjamin et al. (1994) claim that IDEFo and IDEF3 play complementaryroles as support tools for BPIL IDEF3 is claimedto be a 'suitable vehicle for quantitative analysis'whereas the 'more abstract representationalapparatus (sic) provided in IDEFo' is claimed to be a 'powerful tool for conceptualdesign and analysisactivities'. To summarisethe key contrast betweenthe two methodsIDEFo is describedas a methodthat focuseson what happensin an organisationand IDEF3 can describehow things work.
IIDEF3 is also considereda key tool to support BPR by Gregory (1994) who advocatesits use in a twelve step methodology supported at each stage by a combination of IIDEFo, IIDEF,x and 11DEF3.Huff et al. (1991a) proposesa modified version of IDEF3 for BPR to 'simplify its use for novice modellersand uninitiated processowners (sic)'. The modification involves removing synchronous/asynchronousidentifiers associatedwith junction symbols. In a South African defenceindustry application Goosen(1994) explainsthe use of IDEF3 to documentlogistics proceduresand uses 'the repair of componentsin SAAF repair facilities' as an example. A single diagram together with a text specification is used to define the logical sequenceof processesnecessary to carry out 'component repairs'. A modified elaboration document based on a table listing the sequenceof processes(UOBs) with the staff responsibleand the next step in the sequencefor eachprocess is also shown. Much of the researchassociated with the application of IDEF3 is concernedwith process 'metrics', using the diagramsas a meansof describingand analysingquantitative information about the
AT&T ISTEL Ltd. (Redditch,Worcestershire, UK).
39 subject under review. The description of a cost benefit analysisapplication by Benjamin et al. (1993) is an examplewhere they propose an approachto estimatethe cost of processes. using PFNs where 'the cost of a seriesof UOBs is simply the costs of the individual UOBs but processesafter exclusiveOR fan out junctions require the use of probabilitiesto establish average,optimistic and pessimisticprocess costs. The approachinvolves building an IDEF3 processdescription, establishingprocess costs and 'rolling (a sequenceof processes)up to a more abstract,less detailedIIDEFo activity'.
3.5 A SURVEY OF FDEFoPRACTMONERS
0 An important finding from the analysisof publishedresearch, carried out from 1990 to mid 1994 involving over one hundredand fifty publications,was that there were few examplesof industrial applicationsof IODEFomodelling. The majority of the publications being wholly or partially academicbased. In responseto the finding the major UK software tool vendo? was approached and agreed to support an analysis of his UK customer database. The initiative was designedto give an insight into a group of IDEFo users and also to investigate the value of carrying out a wide ranging survey with a larger target population involving a larger range of modelling methods. The findings from this survey were presentedto the EPSRC Workshop on StructuredMethods and Tools in November 1995 and the responseof the workshop to the issuesraised is reported by Little (1995).
3.5.1 Questionnaire design
The design of the questionnairedrew on the work of Oppenheirn(1970) and Crimp (1981), it considered the topics to be included and omitted and a depth of investigation that balance deterring information. blank attempted to - respondentswith capturing usefid A modelling methods questionnaire is shown in Appendix 8, - 'tick boxes' were used to maximise returns and the questionnairewas restricted to a single sided A4 card with a
McroMatch Ltd. - At the time (late 1994) the only UK vendor actively promoting IDEFo softwaretools.
40 BusinessReply label on the reverse. The questionnairewas included in a commercialmail- shot carried out by NlicroMatch Ltd. Twenty three responses were received which representsapproximately 20% of the questionnairessent to the one hundred and eighteen Design/IDEF usersin the UK.
Seventeenquestions were used, Open questions (1,3,4,6,8) together with Classification questions, (2 and 5) Rating questions(16,17) and closed questions (11,13). The use of Open ended questionswas restricted to investigatingthe nature of the models used (questions 7,9,10,12,14,15). In this case the potentiaUy wide range of responses precludedthe tick box approach.
Five aspects of IDEFo use were addressed:(i) the type of organisation using IDEFo modeffing, (H) the modelling practitionerswithin the organisation, (iii) the application areas for modelling methods (iv) the nature of the IDEFOmodels used and (v) a users view of IDEFo.
3.5.2 Results of the survey
I Mncipal activity Number of % of respondents respondents University 6 26 Defence 3 13
Banldng,financial, Business services 4 17.5 Manufacturing(various) 3 13 Govermnent.department 2 8.5 Computerservices 2 8.5 Software& systemsconsultancy 2 8.5 Retaildistribution 1 4.5
Table 3.1 Principal activities againstnumber of respondents.
41 The total number of modelling practitioners in the survey lies between The number 145 and 301 of organisations
4 to 10 llto20 2lto4O- 40+
The numberof staff using modelling in eachorganisation
Figure 3.1 The numberof staff using modefling and number of organisations
IWmaryJobfunctions of Number of % of respondents respondents respondents Businessanalysis 8 34.5 Consultancy 5 21.5 . Research 5 21.5 Teaching 2 8.5
Technical/customersupport 1 4.5 Computer systemsdevelopment 1 4.5 Manufacturing operations 4.5
Table 3.2 Prharyjob functions againstnumbers of respondents
42 Requirements Process Info. system Organisational Research Others Capture Redesign Design Change
IDEFo 12 16 12 6 5 IDEF, 2
IDEFI.
IDEF3 2 2
CORE
GRAI
SADT 2 2 3 2 SSADM 3 Statecharts 3 3
'Petri Nets Checkland 2
DFD 9 6 9 1 Simulation 2 6 3 Others:
Information Engineering methodology Flowcharts
Table 3.3 A Modelling method/Applicationmatrix
43 Subjectof last model Size(no of diagrams)
Mediumscale capacity analysis 16 Combinedrisk/business model 200 Defenceequipment acquisition & support 28+ Customerservices 7 Manufacturingsystem 50 Productdata management system 15 New businessprocessing 7 IS stafftraining and development function 25 No title 47 Structuralsteelwork 30 Softwaredevelopment project 10 Clinicaltrials datacapture and reporting 5 Banking 150 Centralquality processes 6* Utility-customerbilling 16 Orderfulfilment process 40 Engineeringprocess 100 Chemicalprocess development 10 Informationtechnology 50 Aircraft maintenancecosts 8
Table 3.4 The subjectand size of IDEFOmodels
44 Primaryjobfunctions "y IDEFo was selectedas a modelling tool
Businessanalysis Internationallyrecognised, Good at functional analysis,Recommended, Abstraction and Control flow, Found to be the most useful for our purposeat the time, Looked useful Consultancy Widely used in industry in BPR programmes, Ilierarchical decomposition,Rigour,
Robustnessof methodology,Previous favourableexperience, Fit for purpose Research Familiar, Easeof use, Industry standard, Requirementof STEP, Simpleto use, Ideal communicationtool Teaching Popular, ICOM roles, Splits and Joins Technical/customer Recommended+ successfultrial support Computer systems Suitedto businessprocesses development
Manufacturing operations Recommendedby consultants
Table3.5 The relationshipbetween primary job function and the reasonsfor selecting IIDEFo.
45 The benefitsof using Low Medium 11igh. Excellent IDEFo
Easeof modelling 2 2 12 5
Analytical capability 7 6 7
Usefulnessof 3 14 3 finished model I I
Table 3.6 A gradeduser view of the benefitsof someaspects of IDEFo modeRing
The benefitsof using Low Medium 11igh Excellent Design ADEF
Quality of graphics 1 6 9 3
Ease of use 1 7 9 2
Glossarysupport 5 6 6 1
Parent/childlinkage 2 12 4
Table 3.7 A gradeduser view of the benefitsof someaspects of using the Design/IDEF tool
3.5.3 An analysis of survey results
(i) The Type of organisationusing IIDEFomodelling. Table 3.1 shows that over 50% of the respondentswere in the university, defence or banking, finance and businesssectors. The six universities and one practitioner in banking, financial and businessservices have researchor teaching as a primary job function. Three responseswere received from the manufacturing sector. The responsesto (H) and (iii) however indicate, that although few manufacturingcompanies appear to be using modelling
46 directly, modelling is being carried out in manufacturingcompanies through businessanalysis or consultancy.
(ii) Modelling practitionerswithin the organisation. Table 3.2 and figure 3.2 indicatethat the survey responseswere from a modelling population of between 145 and 301 practitioners or users. Over 55% of respondentscite Business analysisor consultancyas their job functions and only one respondentis directly involved in manufacturingoperations.
(iii) The applicationareas for a rangeof modeffingmethods. The responsesto question8 give an indication of the application areasfor IDEFo modelsand are collated in a modelling method/applicationmatrix shown in table 3.3. Numbers in cells representthe number of practitioners using a particular modelling method, for a particular application area. Cells with five or more occurrencesare -shown in bold. All respondents use IDEFo in at least one application area. For information systemdesign and requirements capture DFI)s are-usedwidely in addition to IDEFo. Interestingly process redesign is the most widely applied areafor IIDEFowith information systemdesign and requirementscapture almost equally important. In contrast for DFDs the ranldng is reversed,information system design and requirements capture are more widely used application areas than process redesign.
The range of software tools usedby practitioners in addition to the DesignIDEF (IDEFo) tool includes:
ABC Flowcharter -Flowcharts VISIO -:Windows graphics LOTU9 -Spreadsheet " ERWINAERX AIDEFo& I1DEFx tool " IThink -BPR tool " RADitor -BPR tool " SystemArchitect -Casetool " Bachmananalyst -Casetool " Exellerat'or -Casetool " PROSIM AIDED tool
47 (iv) The nature of the IDEFOmodels used. Over 90% of respondentsused both AS-IS and TO-BE models and in terms of the type of models used the responsesto question 9 and 10 provide a snapshotof the specific subject and size of models last completed (late 1994) by practitioners. Responsesare collated in Table 3.4 and the subjectsof the models are given verbatim. Of the models tabulated 50% used 30 diagramsor less and only three respondentsproduced models having 100 or more diagrams. Six respondentsused ten or less diagramsfor their model. In 91% of casesthe models were used by staff in addition to the model builder inside the organisation. Of the model descriptionscited over 30% are manufacturing related and the two largest models described(in terms of numberof diagrams)are for businessand banking.
(y) a usersview of the IDEFo method. Question 12 is an open endedquestion designed to indicate the reasonsbehind the selection of II)EFo as a modelling method. A review of the responsesrelated to the primary job functions of respondents is given in table 3.5. It was notable that not all returned questionnairesincluded a responseto this question and the open endedapproach produced a range of responses that included methodological reasons for adopting the method, managementreasons and reasonsbased on generalperceptions. The major usersof 11DEFoin Business analysis, Consultancyand Research show a consensusin citing, 'Internationally recognised', 'Widely used in industry' and 'Industry standard' as reasonsfor adopting the method. To establish an understandingof how users perceive the use of DDEF three questionswere selected: to indicate if the method was easyto use, to establisha view on its analytical capabilitiesand to assessthe usefulnessof finished models. Table 3.6 is a matrix summarisingthe responses,numbers in cells representthe number of occurrencesof the cell being selectedby respondents. Easeof use and usefulnessof finished models is rated highly. Analytical capability is however not rated as highly. -Usersrating of the Design/IDEF tool is high albeit with somenegative rating of its Glossarysupport:
A speculativesnapshot of IDEF modelling in the UK manufacturing, based on the limited number of responsesto the survey, indicates that a modelling practitioner is a consultant. He/she usesother modelling methodsin addition to IDEFo, most likely DFDs or Simulation and other modelling software tools are used. Models are likely to comprise of less than 30
48 diagramsand are used by other membersof staff in his/her organisation. The Design/IDEF tool is Wghly regardedand IDEFo was selectedbecause it was consideredto be a standard method.
In termsof the objectivesof this researchthe surveyhas contributedto understandingthe use of modellingmethods particularly IIDEFo in manufacturing. It provides a strong indicationthat manufacturingcompanies themselves are not using IIDEFowidely and re- enforcesthe findingsdiscussed in section3.6, that a significantuse of IDEFoin the UK is throughUniversities and research. The surveyresults have made the major contributionto the designof a largernational survey of manufacturingenterprises in the UK (Small,Baines and Colquhoun,1996) that seeksto investigateother modellingmethods in use and to providean insightinto thoseenterprises who havenot consideredor haverejected the useof modellinggenerally. The larger survey is not reportedin this thesis.
3.6 CONCLUSIONS
When the initial state of the art review of IIDEFowas carried out (Colquhoun et al., 1993) a majority of publications were concerned with category 2- performance evaluations or comparisons. There is evidence now that the emphasis of publications has changed, a majority of later publications are in categories 3- proposals for enhancementsand 4- specific applicationsof the method. This reflects the maturity of the method, that its use is establishedand well understood and that research is focusing on perceived weaknesses, developmentand on applieduse of the method.
The basic concepts are now well recognised as evidenced in both the amount of related published work and in the general use of the method to represent the concepts of manufacturingsystems. In those publicationsproviding descriptionsand reviews of the basic method, most introduce the method and its potential applicationsand restrid descriptionsto activity diagramconstruction. Few authorsdeal with the broader aspectsof the method such as model structure, author-readercycle and the interview methodology.
49 The majority of reported work has taken place in the USA as a result of the impetus provided by the US Air Force. In discussingits use in the USA Stephansand Fox (1987) report that 'the method is used primarily to review government contracts and there is relatively little data concerning its application outside the defence or government environment',the claim is supportedby this review. Few completeIDEFo models are in the public domain however a comprehensiveanalysis of aerospacemanufacturing has been provided by CAM-I (1980) in their model 'Get and use aerospaceproduct. Use of the four categoriesin this analysishas revealed that applications of the method are the least well documented and that few detailed applications of IDEFO have been reported by UK enterprises. The survey in section 3.5 indicates that the method is being used more widely than the literature survey would suggestthough the evidenceof the survey re-enforcesthe argument that non-academicmanufacturing applications in the UK are very limited. The emphasisis on a largely academic/researchapproach to the subject and could be an with indication that the rich picture that an IIDEFomodel can provide may be in conflict the confidentialnature of industrially basedmodels.
Early researchersraised the problem of a lack of computer tools to support the method for examplein 1983 Yadav (1983), using SADT, found that 'for large and complex systemsthe lack of automation makes an SADT model vulnerable to inconsistencyand incompatibility' Other authors refer to the large number of diagramsand amount of documentationrequired for an IIDEFOmodel and seethe tedious nature of diagram production as a drawback to its use. The proliferation of IIDEFo computer tools in the 1990's has largely overcome the problemsof diagramproduction and model administration.
However the building of an IIDEFo model of a manufacturing system can be a difficult processdemanding a great deal of time and care, manufacturingsystems are truly complex with many formal and informal relationships occurring between sub-systemsand their representation tests any modelling method. Wu (1992, p.330) observes that 'actual operations are likely to be "greased" by informal systemsthat often provide methods of overcoming short-term problemsand occasionallyallow an unworkable operating procedure
50 to work and so it is essentialthey are consideredin the analysis'. There is a consensusthat IDEFo is powerful enough to stand up to modelling both formal and informal functional relationshipsand that experiencedpractitioners of the method can provide useful and largely accuratemodels. Few authors are wholly critical of the method and most support the basic tenet that the method provides a meansof understandingthe complex interaction of men, machinesand information, and a meansof communicatingthat understandingto others. It is evident however that most authors agree no one method can model the functional, information, dynamic and decision aspectsof systems. It is therefore surprising that the interfacebetween 11DEFo and other IDEF methodshave not beenwidely discussed.Goldman and Cullinane (1987) identify.a need for a linicing methodology to develop IDEFI models from IDEFo models and some other authors have discussed IDEF method integration (Godwin et al., 1988, Schneider, 1994)
Amongst the criticisms of themethod Shunk et al.. (1986) claims that 'diagrams often go unused' and Harrison et al. (1988) claims that the effort in diagram production involves a significant amount of work. Busby and Williams (1993) claim that 'IIDEFocannot represent in any profound way the mannerin which people take decisions'. Goldman and Cullinane (1987) also suggestthat limitations of the method are related to the lack of decisionrules for model decompositionand the lack of an explicit link between information flows and claim that decomposition is frequently misunderstood by users. 'The most frequently cited shortcomingof the methodis the lack of a 'time' dimensionto describeactivities.
IIDEFOhas been designedto model manufacturingand no other modelling method claims to provide the samefunctional analysiscapability, it is accessible,in the public domain and has strong software tool support with a large body of knowledge and a proven capability in the US defencemanufacturing arena.
51 CHAPTER 4
A COMPARATIVE ANALYSIS OF MANUFACTURING ENTERPRISE SYSTEMS ANALYSIS AND DESIGN METHODS
4.1 INTRODUCTION
This chapterreports on an analysisof methodsused to model manufacturingenterprises. It focuseson methodsthat predominantlyuse models basedon diagramsand sets of diagrams representingvarious aspectsof manufacturingenterprises. The descriptivemethods, diagram syntaxand model structure usedin graphical presentationare also analysed.
The analysisre-enforces a generally acceptedview (Jones et al., 1991, Roboam and Pun, 1989) that no single method can be used to model the full compleidty of a manufacturing enterprise. The methodologies that underlie the various methods were also reviewed. Shortcomings derived from analysing the capability of each method provide a basis for recommendinghow aspectsof several approachescould be integrated to provide a hybrid approachrobust enoughto model a broader range of manufacturingenterprise characteristics than has previouslybeen the case.
This chapterextends the work publishedby Colquhounand Baines(1993b).
52 4.2 CONTEXT FOR A COMPARATIVE ANALYSIS
Definitions:
Modelling method -A manual or computer prepared graphical and text descriptionbased on a-formal documentedsyntax such as IDEFo or other structuredapproaches.
Methodology -A method or collection of methods or computer tools whose use is governedby a proceduresuperimposed on the methodor coHection. Tool - (as in CASE tool) a software systemdesigned to support the application of a modelling method or methodology.
Widely recognisedfonnal' graphical modelling methods for the representationof concepts, design proposals and the transfer of information are routinely used disciplines such as architecture, engineeringdesign and software engineering. In the emerging discipline of manufacturing systems design and analysis, engineers are dealing with complex system characteristicsthat range from human constraints,to material flow. In this environment a clear consensuson the application of the many formal modelling methods that have been proposedhas yet to emerge. A generalcommitment to the use and standardisationof formal methodshowever is evident in the scaleof researchinitiatives suchas CIM-OSA, launchedin 1984 involving twenty one companies from seven European countries (Jorysz, and Vernadat, 1990a, b, Klittich, 1990, Klittich, 1995) and the US Air Force Information Integration for ConcurrentEngineering (HCE) program (Painter, 1992) that has led to the developmentof the IDEF3 to IDEF14 family of methods.
Modelling methods used in manufacturing systems design and analysis are an aspect of problem-solving that have diverse roots in Operations Research, Software Design and General Systems Research, three broad areas that pre-date the emergence of manufacturing systems engineering as a recognised discipline. Modelling methods can combine both quantitative and descriptive aspectsof a situation and can be either static or dynamic and can describe an existing situation or conceptualisefuture situations. There
1Tomml' refers to the a standardisedset of rules and constructsfor graphical presentation.
53 are no distinct categories of modelling methods for manufacturing systems however the two broad domains of 'Hard' and 'Soft' systems approaches (Wu, 1994, Cavaleri and Obloj, 1993,Boardman, 1990) have become accepted.
Hard systemsapproaches are based on rational decision making to solve problems that are well defined and that can be largely described quantitatively and hence solved mathematically. Wilson (1984) proposes four categories for mathematical analytical modelling, steady state and dynamic (modelling over time) each of which can be either deterministic, an algorithmic approachthat assumesall the relevant system parametersare known, or non deterministic demandingthe use of statistics that assumea range of values can describe a particular situation. The goal of hard systems approachesis to identify alternative solutionsto a problem which can subsequentlybe evaluatedagainst some criteria such as risk, costs or benefits.
In contrast soft systems approachesare proposed to deal with ill-structured problems without well defined objectives or a commonly agreed problem definition and with incompleteor few quantitativemeasures of performanceor behaviour. Real world problems that Boardman (1990) argues 'are more complex than hard systemsmethodology admits' becausethey involve, or have arisenbecause of, the different perceptionsof those involved. The contrast to hard systemsapproaches is re-enforcedby soft systemgoals that are 'more concernedwith the orchestration of debate centred on perceived problematical situations' (Checkland, 1989). In a soft systemsapproach there are no permanentoptimised solutions to a problem situation but the tenet that a continuousseries of improvementsare possible.
Hard systemsapproaches for manufacturing enterprise systems assumethat all necessary information will be availableand that problem solutions will be implementedas predicted by a mathematicalmodel using constraintsto meet certain criteria. The approachis in conflict with many aspectsof manufacturingenterprises which are subject to the control of humans and can be dynamic,uncertain, and in some conditions chaotic. It is at a tactical level that hard systemsmethods tend to be applied 'becausemany operational goals have been refined by the time they reachthis level and can thus be more readily measured'(Cavaleri and Obloi, 1993).
54 Figure 4.1 is an overview of a range of systemsmodelling approaches in the context of 'hard' and 'soft' approaches.The spectrumof modellingapproaches positions a representativeselection of methodsthat havebeen proposed for modelling aspectsof the manufacturingdomain.
SYSTEMS
EDEF2 IDEF4 ""ýIDEFIX ,,, IDER GRAI HA" \, SOFT IDEF3 SADT/ Pet,rli D ataFlow IDEFO Net I: gmms vsm CPN I. S. Checklands Stochastic Soft Systems simulation, General 0. P- Mamifacturing I&thmatical Systems modelling enterprise systems modelling Origin of methods O.R. - Operationalresearch I. S. - InformationSystems CPN - Critical PathNetwork VSM - Viable SystemsModel
Figure 4.1 The modeflingmethods spectrum
The 113EFmethods have substantiallyinfluenced this researchand have provided a reference point (rather than a benchmark)for a review of someof the modelling approachesreferred to in the modeflingmethods spectrum of figure 4.1
4.3 SOVr SYSTEMS MODELLING METHODS
Soft systems modelling methods rely on viewing systems through the perceptions of participants in the systems. The approachacknowledges that there may be no single wen defined problem in a complex dynamic system and that it is only by exposing, discussing and refining understandingof a systemthat improvementscan be made. Two clearly 'soft' methodologies have been applied in manufacturing enterprise modelling, the Checkland
55 Soft SystemsMethodology (discussedin appendix 4.1) and the Viable SystemsModel (VSM) (discussedin appendix4.2).
VSM has been describedas a theory of organisation, and Espejo (1989a) advocatesthe approachas a 'paradigm for problem solving' and a tool 'to diagnosethe effectivenessof an organisationsstructure'. It is certainly more than an analysismethodology and shares some structural characteristicswith the Checkland approach. The directed modelling methodology in terms of the systems; (operating core, control and co-ordinating mechanism,internal information systems,external information systemsand policy maldng) has parallelswith the CATWOE of Checkland. VSM uses the principle of 'recursion' to examineaspects of a systemin greater detail and in a structured manner that reflects the decompositionapproach of hierarchicalmodelling methods.
The essenceof the method does not lie in graphical representationbut in the use of the model to guide the intellect of the analystin dealingwith a problem. The potential dMiculty of routine applicationin manufacturingenterprises lies in the intellectualrigour necessaryto apply VSM concepts.
The structured diagramming'of 'harder' approachesdemands more sophisticatedsymbols and representationsin order to be able to map the problem to the model. Whereasthe VSM and Checkandtend toward mappingthe problem to the concept. Anderton (1989) draws an interestingly similar contrast from a different viewpoint between 'harder' approachesof systems engineersand the 'softer' VSM and Checkland SSM highlighting the rationale behindthe, apparently,casual sketch approachused to presentsoft systemmodels.
'They (Beer and Checkland)insist their freehand drafts are reproducedunchanged with the slightly wiggly lines retained. They prefer cloud shapesto boxes. Intuitively they feel that somethingof the richnessof their propositions, of the organic origins of what they seek to represent,is lost by a conversioninto perfect grids, squaresand other geometric artefacts.'
The GRAI (Graphei R6sultatset Activites Interli6s) methodologyhas been influencedby IS and general systemsapproaches and is at the 'softer' end of the modelling spectrum. The
56 method lies in the samearea of the modelling spectrum(shown in figure 4.1) as IDEFo. It is focused on representingreal aspectsof a manufacturingsystem but can representconcepts and can deal with subjectivityin its graphical representation.A review of the methodologyis presentedin appendix4.3.
4.4 INFORMATION SYSTEM MODELLING METHODS
Modelling methods associatedwith data modelling and software design He in the centre of the modelling methods spectrum. These modelling methodshave had a major influence on the development of manufacturing enterprise modelling in both proving the value of in the development modelling and of methods and methodologies..
The emphasisof diagrammingmethods such as data flow diagramsand data model diagrams has been to provide a concise description of an existing or a proposed system in order to build software to run or to support the, system. The rationale behind the developmentof such methodshas beento ensurefault free design,to eliminate data redundancy,to minimise software development time and to render large computerisation,projects manageable. Rationalisation or redesign of the system structure was an outcome of the application of computersrather than an end in itself. The recent rise of BusinessProcess Re-engineering (BPR) which according to Gutkowsld (1994) 'was a provocative concept only a few years ago' and the developmentof Computer Aided SystemsEngineering (CASE) tools used for the automation of structured methodologies (Britton and Doake, 1993) has shifted the emphasis to focus on re-organisation of systems before designing and implementing computer systems.
In contrast, the primary focus of manufacturing enterprise systems modelling from its beginningsin OperationalResearch in the 1940's hasbeen problem solving or systemchange to improve performance. Some of the accepted principles of modelling methods have however been developedfrom software design methods such as 'Top Down' programming from the early 1970's and structured design and structured analysis (Yourdon and Constantine, 1978, Yourdon; 1989, DeMarco, 1979). Cavaleri and Obloj (1993) see
57 SystemsAnalysis (in the context of Information Systems)as an approach that integrates quantitative aspects of systemswith organisation theory and that it 'acknowledges that systemsare often pluralistic in that they contain many decisionmakers with diverseand often competing values'. Re-enforcing the argument that information system modelling has characteristicsof both hard and soft systemsapproaches.
An analysisof the range of graphical modelling methodsin the information system domain exposedthe fbHowingbroad categories:
(i) those usedto model data from a logical viewpoint. (H) thoseused to modelprocesses (or activities)and the flow of databetween them. (iH) those usedto model the structure of software.
(i) In this category a variety of similar graphical notations have been used to represent entity-relationship concepts based on relational theory proposed by Chen (1976), Codd (1970) and others. The entity-relationship(E-R) modelling used in the IDEFx method is an example. In basic form Entities (representedby rectanglesin figure 4.2) are 'things' or collections of 'things' which have a relationshipwith each other (representedby the joining line and a verb phrasewritten along the line).
The cardinality of the relationship is describedby symbolson the joining line, in figure 4.4 the entity-relationshipdiagram represents a situation where a customer (the entity customer has data ) (similarly, the attached such as name , addressetc. canplace many orders entity order has attacheddata such as order number,quantity, description etc.). From a normalised entity-relationshipdiagram (or schema)the tables and table structure of a relational database can be established.
58 CUSTOMER
Parent (Independent)enUty
PLACES
I "-- Relationship ORDER CustomernLmnber (FIQ Ordernunber Child Pependent) entity
Figure 4.2 An IDEFIx representationof an entity-relationship
Terrr 3tor
Dat3 stcwe
Figure 4.3 The elementsof a DeMarco Data Flow Diagram (DFD)
(H) Various diagrammingnotations have been proposed to representthe flow of data in a software systemand are largely basedon DeMarco's (1978) data flow diagramning method. A Data Flow Diagram (DFD) depicts processesand the flow of data betweenprocesses as a graphicalnetwork to identify the requirementsof software systems. Figure 4.3 is an example of a data flow diagramusing DeMarco's graphical symbols.
59 A DFD has four graphical elements:(1) Processes-a processinvolves data transformation (for example,using arithmetic or logic operations) and is representedby circles or round cornered rectangleswith a concise description of the process inside. (2) Data flaws are arrows tracing the path and direction of data flow from processto process.(3) Data stores or files are representedby horizontal lines with the name of the data store written above. The direction of arrows entering or leaving a store indicates a 'read' or 'write' operation. (4) Terminators are squaresindicate the originator or receiver of data and are shown on the periphery of a diagram.
Other DFD approaches include Gane and Sarson (1977) and the Structured Systems Analysis and Design Methodology (SSADM) Longworth and Nicholls (1987). Figure 4.4 shows the DFD notation from SSADM the core elementsare the same as DeMarco but different symbolsare used. Data Flow Diagrams can be used to model high level and lower level, detailed views of a system, what takes place in a higher level 'process' box can be explodedinto a lower level diagram,the decompositionof diagramsis termed 'levelling'.
Data source
Data Rem
Process --Lj Data store
Data Rom
CData d
Figure 4.4 The graphical elementsof an SSADM Data Flow Diagram
Some authors have proposed modifications to the DFD; Ward (1986) claims that DFD models have not provided a comprehensiveway to represent timing, control and data transformation and proposes extensionsto model structure to overcome the problem. To
60 use DFDs in manufacturing MacIntosh (1993) proposesa structured integration of DFDs and GRAI grids. He usesDFDs to describeinformation flow and builds a GRAI grid for 'to act as a summary' for each Data Flow Diagram and provide a picture of the decisions governinginformation flow.
(iii) Early approachesto modelling program structurewere basedon flowcharts but the lack of functional decomposition and the emphasison detailed logical flow meant they were of less increased. (Nassi use -as program size and complexity Nassi-Shneiderman and Schneiderman,1973) diagramsattempt to overcomethe disadvantagesof conventionalflow chartsusing a nesting syntax which is claimedto be easilytransferable to structured program code. Warnier-Orr diagrams (Warnier, 1981) use a hierarchical text list that shows the logical relationshipsbetween processes (processes in this casebeing routines or sub-routines in a program). Michael Jackson diagrams and Ilierarchical Input-Process-OutputOHPO) diagrams (Martin and McClure, 1985) have also been proposed as graphical methods to model program structure.
The three categories discussed: modelling data from a logical viewpoint, modelling processes(or activities) and the flow of data betweenthem and modelling the structure of software are steps in complete methodologiesfor software development. Three such IS derived methodologies have been widely discussed in the context of modelling for manufacturingsystems:
The Structured Systems Analysis and Design Methodology (SSADM) (see appendix 4.4).
* The Nijssen Information AnalysisMethod (NIAM) (seeappendix 4.5).
a The COntroUedRequirements Expression (CORE) method (seeappendix 4.6).
Over the last decadea wide range of support tools have been developed. Computer Aided Software Engineering (CASE) tools has become the generic term for such systems. Somerville (1992) proposes that there are several hundred commerciallyavailable systems.
61 The scope of CASE tools for software developmentincludes system specification, design, implementationand validation and their functionality can include: modelling and simulation, languageprocessing, method support, prototyping, documentpreparation, project control.
4.5 'IS' MODELLING METHODS IN THE CONTEXT OF MANUFACTURING SYSTEMS
A range of IS modelling approacheshave been proposedto model manufacturingenterprise systemsand in many caseshave been comparedto 11DEFo(Chapter 3 section 3.3.2 discusses comparisonswith DDEFoin detail). This section discussesspecific IS modelling approaches, which, the researchto date has established,have the greatest potential to contribute to modelling in manufacturing.
DFDs are amongst the most popular modelling methods used in manufacturing systems applications. The modelling principles of DFDs are very close to EDEFOand DeMarco, (1979) argues that 'an SADT diagram is a'sfight variation' of a DFD. A complete DFD model involves decomposition of a single top level diagram called the 'context diagram' which defines the scope of the model. Just as in SADT and IDEFo, rules based on 'balancing' parent and child diagrams provide structured decomposition of higher level diagrams.
In contrast to a DFD an DDEFOmodel diagramis constrainedby a more rigorous set of rules however the conceptsand model building processof DFDs are analogousto those of IDEFO. Two key differencesbetween the methodsare the specific data flow focus of a DFD and the identification of 'Controls' in IDEFo. The arrow structure of IDEFo is also designed to representobjects in addition to data or information. Used as originally intended a DFD is a 'harder' approachthan IDEFOconceived to define software requirementsaccurately and as such would not be intended to describe concepts or provide a subjective description of a situation.
62 The focus of data flow diagramson processes has lead to its application in manufacturing enterprise systems modelling. For manufacturing, DFD graphical representationhas no advantageover IDEFo, it cannot describe any additional characteristicsof a manufacturing enterpriseand IDEFo provides a much richer description. I would suggestthat the simplicity and lack of rigour in diagramformat has had some influence on its adoption. Little training is necessaryto understanda DFD (despite the intellectual effort necessaryto build models) only four symbols are used, no constraints are imposed over diagram format and the diagramscan be understoodintuitively. One important lessonto be learnt from the DeMarco DFD approachis that modelling method complexity must be kept to a minimum. However the approachgenerally does not attempt to describethe time processestake or the logic of processsequence or the conditionsthat govern processsequences.
CORE has been specifically designedfor modelling in manufacturingapplications. Using CORE a processsequence can be describedhowever alternativesequences (or branches)are limited to two formal constructs; 'mutual exclusion' and 'indeterminate order of actions'. On the other hand CORE has the ability to describefive different types of action iteration conditions. The focus of the method is on the descriptionof data in relation to actions rather than on the relationshipbetween actions. A CORE model involving the six types of diagram with sophisticatedmodel syntax and Node notes is an exhaustivemeans of exposing detail data requirementsin a complex 'one of a kind' manufacturingenvironment. A characteristic of both CORE and NLAM is the ability of the methodsto describetypes of data. In thread diagramsCORE can describe; event data, data containing information and control data and distinguishesbetween 'critical' and 'non-critical' data. NLkM makesthe distinction between objectsand data about objects.
4.6 'HARD'SYSTEMS MODELLING METHODS
Hard system modelling methods rely on a reductionist approach, breaking-down problem situationsinto clearly definablepackages where all the parametersrelevant to a problem are known and where statisticalmethods or algorithms can be appliedto predict future eventsor solve problems Cavaleri and Obloj (1993) claim that OR fies at the core of hard systems
63 approachesand that the critical foundation of hard systemsis 'a well defined problem'. 'Hard' graphical methods have been used to model various aspects of manufacturing enterprises. Sauter and Judd (1990) argue that as manufacturing systemsare man made machines(sic) 'they cannot be describedby differential or diffferenceequations, but rather they generateevents which must be controlledand that such systemshave been termed DiscreteEvent DynamicSystems (DEDS)'. This review doesnot considermathematical models suchas Markov chains,queuing theory or inventory models.
The three important 'hard' modelling methods in the context of this work are discussedin appendix4.7,4.8 and 4.9, they are; Discrete Event Simulation, Petri nets and Critical. Path AnalysiS'(CPA).
4.7 GRAPHICAL REPRESENTATION
Most of the methodsanalysed have been proposedby practitioners as suitable for modelling man, machine,material and information systemsin a manufacturingcontext. Some of the methodsare, in the broadest sense,methodologies. The analysishas however concentrated on the graphical or textual descriptiveaspects of the methodologiesor methods. They have widely dfferin$ applications however the analysis has revealed that they share some fundamentalcommon characteristics:
They all use a graphical descriptive language having a formal syntax and use symbols to representattributes or characteristicsof a systemand the relationshipsbetween symbols,text descriptionsand numbering systems are used in conjunctionto form a completemodel.
The scale or area of their respective diagrams, line length or symbol size do not have any descriptivesignificance.
The relative position of symbolsis usedto describesome system attribute.
64 Arrows or lines are used to describethe relationshipsbetween symbols in terms of sequence or flow.
The principle of abstraction(infonnation 'hiding') is used to exposedifferent levels of detail andthe applicationof abstractionover severaldiagrams is the basisof a completemodel.
Text is used as an additional meansof systemdescription, some methods use formal rules to constrainthe nature of text within models. Checkland(1981) for instancesuggests that the root defHtion of the Checklandsoft systemsmodel shouldbe in the form of 'a structured set of verbs in the imperative mood'. In IDEFo an activity is describedby a 'concise active verb phrase' (Ross et al., 1980).
Recognition of these common characteristicshas provided a framework for the structured comparativeanalysis discussed in section4.8.
4.8 A STRUCTURED COMPARATIVE ANALYSIS
A structured analysis of the characteristics of a representativerange of methods in the context of manufacturing enterprisesis presentedin this section. The methods have been examinedin chapters2,3,4 and appendix 4 and selectedbecause they have been used to model aspectsof manufacturingsystems. To provide a structure for the analysisit has been necessaryto identify modelling parameters that would be useful when attempting to representa wide range of manufacturingsystems features. In total 14 modelling parameters have been identified and presentedin a comparativetable (table 4.1 a, b, c) that describes characteristicsassociated with each method considered. The table also highlights, in italics, the graphical symbols or representationadopted by the method to describe the states or characteristicsof a system.
Following is an explanationof the modelling parametersused in Table 4.1 (a, b, c).
1. Model state:Three states have been identified:
65 * An 'AS IS' model representsa situation as it exists in practice, defined in company specifictemiinology. * 'Generic' describes a model that represents the generally accepted or intrinsic activities that exist within a defined process or system and wiU represent all occuffencesof those activities (Colquhounand Baines, 1991). * 'Planning' describesa model that can be used to expressa design or to conceive or propose a future state of a system(an 'TO BE' state),
'2. Basic elements: Are the aspectsof a systemthat provide the focus of a method and basis of the diagramsproduced. The symbolsused to representthese basic elementsare the key elementsof a model diagram.
3. Link between basic elements: This modelling parameter refers to those aspects of a system that relate basic elements (parameter 2) and are shown diagrarnatically as symbolsthat link basic elementsymbols.
4. Nature of the link between basic elements: In most casesthe nature of the link between the basicelements will be physicalsuch as material,information or data. However they could also be linked by concepts, for example by a 'decision to proceed' or 'designideas'.
5. Basic element dependency: Two conditions have been considered either the basic elements'will be non-conditionalin which case the method is not capableof modellingthe conditionsfor a basicelement to occur,proceed or exist. Whereasa conditional relationshipis one where the method is capableof modelling the conditionsfor a basicelement to occur.
6. Basic element time span: A method's ability to model the time spanof basic elementsor betweenbasic elements,two stateshave been considered; (i) 'intrinsic' meansthat time is an essentialelement of the method and is representedusing graphics or text
66 and (H) 'not represented'means that the time spanof basic elementsis not considered in the modeffingmethod.
7. Nature of time span: This is closely related to parameter6 and considerswhether the modelcan support deterministic and/or stochastic times.
ý. Representation of concurrent elements: This simply considerswhether the method can modelbasic elements or linldngelements that occurconcurrently.
9. Representation of dependent elements: Similarly this parameterconsiders whether the method can model any dependencyor sequenceof basic elementsor linldng elements.
10. Representationof resourcerequirements: Is the ability to describethe associated resources(in terms of people, machinesfinance etc.) necessaryfor carry out basic or linIdng elements.
11. Representation of model hierarchy: This refers to the ability to provide a structured hierarchyof diagrams.
12. Supporting information: This considers the formal rules associated with text descriptionsnecessary to support the basic diagrammingmethod.
13. Model validation: This considersthe formal meansof model validation associatedwith a method.
14. Complete model context: This relates the ability to establish the model in a wider context.
67 rA Gn rA m Z ,Wý c3 124 ý,0
r4 W
lo. % .0 en as GO2
Z to; 43
< Z
,a
0 1
', 02 W Z Z d
Z El IR
z2
2' EA
Z
Z Gof 0 2 r. C.2 23 ' i 22qZ lz
Ao W ýý Z 2
220. r4 1
21 ýz0 o m Z Z <
A
0
r4 en 4e 4-4 \O L-PN-4
68 rn rA r02 ý. . Q op ý 0 Z Z Z Z0
--ez ý3 r4 Zmý, Z 2, ro,- o " Z u
rA rn c> C, u2 Zý Z Z
rn
1U4 Cd
CD
04
r2N-4 Z zo 4-tE 0 ýzusg
UGo
ce 22 .0 E4 RA eý
ý iz 1Z CK2 9ý u F-4
Z Z 0 Z Z ýz1 ied
M0ýi 11l 0 ' ze oo 1 11w44 __j
69 c02 GO2 Zk GO2 44
ej 1,02 l's,2
Z Z W
t),Cd
,a 0 Z Z Ei bo A 1 rA W 1 ý.0 u 4 0 ý, l' E ý 2 9 N-4 zý 9K etý3 0 2ak : k -, 0 Z Z
ri Q- vi 1 u2 M ýý zýZ
C) zm .,2 - 21 c9
ZO ý5p Q-iä <
0 JL. 22 X 0. 0 1 1 1 -- I p04 1r4 1en In. I P%.4 e%4 p44 p*-q
70 Table 4.1 provides a structuredcomparison of the graphicalsyntax employedin the methods. The range of methodsreviewed have widely differing approachesto representingaspects of manufacturing systemsand all have been designedto model manufacturing systemsfrom different perspectives. However all, with the exception of DFDs, have a form of 'transformation' event or activity as the basic elementof a diagram. In all methodsthe basic elementsare linked diagramaticallyby some other characteristicof a system,in the case of DFDs the 'transformation' is the linking element. Labelled arrows and boxes are commonly used graphical symbols. Most of the methods reviewed support the diagrams with a methodology that includesvalidation and a meansof describinga context for models and in somecases a formalisedprocedure for presentingsupporting information is used.
The comparison indicates that for IIDEFomodelling parameter6 and 7 are not represented and 8 and 9 are not intrinsic. Parameters6 and 7 relate to the ability to model 'time' and 8 and 9 relate to the ability to describe aspects of a sequenceof activities. Parameter 5 indicatesthat IDEFo is not capableof modelling the conditionsfor a basic element(activities) to occur, proceed or exist. Three methods, IIDEF3, GRAI and CPA, however are able to represent aspectsof parameters5,6,7,8, and 9 but of the three only IDEF3 has a similar 'basic element' modelling parameterto IIDEFo.
4.9 CONCLUSIONS
Aff the modeffing methods are specific and focused on some aspect of a system. They all omit non-essentialdetails and. concentrate on someparts of reality at the expenseof others accordingto the aimsand viewpoint of the analyst.In a complexmanufacturing enterprise a singlemodelling method may not be sufficientlybroad to capturethe complexityand a series of modelling approachesmay be necessaryhowever none of the methods consideredin the comparativeanalysis claim any syntacticalor methodologicallinks with other methods.
Arguably BDEFohas been the most favoured approachto provide static graphical models of manufacturingsystems. (A recent study in UK manufacturingreported by Baines, Small and Colquhoun (1996) supportsthe claim but also revealsthat DFDs are almost as widely used.)
71 The comparative analysisre-enforces the argumentthat IDEFOprovides a modelling ability over a broad range of manufacturing enterprise characteristics. It also revealed some limitations of IDEFo when modelling fairly commonsituations in manufacturingsystems such as; - " the absenceof a facility to modeltime " the lack of a facility to modelthe orderingof activities(sequence or logic) 4! conditionalrelationship between activities
There have been severalattempts to improve IDEFo using diagrammingmethods where the fundamental concepts have been retained yet extended to make them more amenableto modelling systemsin manufacturingenterprises. Mandel (1990), for example,proposes the use of 'ports' on IDEFo diagramsto overcome the problem of identifying the sourcesand destinations of arrows crossing a diagram boundary together with a modified diagram format.- Shunk et al.,. (1986) propose the Triple Diagonal Technique (IDEFo-TD) using a modified IDEFO diagram structure to help distinguish between information, control and material flow (other proposalsto modify the basic IDEFo method are discussedin section 3.3.3). There is no evidencethat these modified IDEFo methodshave been widely adopted, supporting the caseadopted in this researchfor combiningor integrating establishedmethods to expandmodelling capabilityrather than modifying robust well understoodprinciples.
It could not be argued that the other methodsreviewed in this paper do not have their own special advantageswhen they are applied to the modelling applicationsfor which they were designed. Some of the methodsreviewed are not structured analysistools in the strict sense of the term but they do exhibit many individual modelling characteristicsthat if used in a structured framework could have mainstream applications in modelling systems in manufacturingenterprises.
The modelling characteristicsthat are candidatesto be incorporatedinto any new approachare a representation of basic element'dependency, the ability to describe time and a facility to model the logic of activities. This meansthat an enhancedIDEFO method or a combinationof methods could have the additional capability to model concurrent and simultaneousactivities, the state of constraint dependencyand the time span of activities or linIdng activities using a
72 commonly accepted set of diagramming symbols comprising of rectangles, arrows, tokens, etc..
The analysisinforms an approachtaken to develop a new method using IDEFo as a starting point and adding to it, some of the modelling characteristicsidentified in the reviewed methods that would have a strong application in modelling manufacturing systems. In any new method it would also be useful to adopt the same diagramming symbols (those highlighted in italics in table 4.1 a, b, c) from the sourcemethods because their use is already understood by analysts and engineersand this will help to increasethe acceptability and widespreaduse of any improved method. The methodologyproposed in chapter 6 takes this approachby correlating IIDEFOand IDEF3 PFNs to provide modelling capability in a larger range of modelling parametersthan IDEFo and IDEF3 modelsalone.
73 CHAPTER5
REFERENCE STRUCTURES FOR MANUFACTURING ENTERPRISE MODELLING
5.1 INTRODUCTION
This chapter introduces the concepts of reference models and system architectures for manufacturing enterprisemodelling. The use of a generic model of process planning is explained and discussedin the context of reference models for manufacturing. This aspect of the research is based on the work published by Colquhoun et al. (1989), Colquhoun and Baines (1991), Baines and Colquhoun (1991). A proposal for a manufacturingenterprise modelling framework is explainedand its potential as a frame of referenceas part of an enterprisemodelling methodologyis discussed.
5.2 MANUFACTURING SYSTEMS ARCHITECTURES
Computer Integrated Manufacturing (CIM) is a manufacturing paradigm, largely developed through the 1980's. It has come to mean the control and execution of all phases of manufacturing from the concept phase through product design and manufacturingplanning to manufactureusing computersas the meansof integration. It was recognisedin the early stagesof CIM evolution that a major impedimentto progress was compatibility and data transmissionbetween the many rapidly developing software systems. Computer Aided Design (CAD) and machinetool control systemsfor example
74 were developedby different vendorsto become'islands of automation', effective as stand alone systemsbut dffficult or inefficient to integrate. One responseof researchbodies was to fund efforts to overcome integration problemsto enablevendors to design CIM systemswithin a common framework. Early examplesinclude the European ESPRIT phaseI-CIM project (Yeomans, 1987b) which provided flowcharts and text descriptions of the generic activities that comprise the machining sector of manufacturing. It is claimedto provide 'a EuropeanCIM architectureagainst which IT vendors could fashion CIM products'. Subsequentlymuch of the ESPRIT phase II research was aimed at developing open-systemsCIM architectureand communicationsto support multi-vendor environments. .
In the USA the ICAM architecture (discussedin section 2.2) was taking a. hierarchical architectureapproach to do the samething. The developmentof the ISO Manufacturing Automation Protocol (MAP) grew from a proposal (International Organisation for Standards,1986) to 'create a multi-dimensional,open ended reference architecture and provide a basis for long-range planning and standardisationthrough the identification of interfaces and their characteristics, electrical, mechanical, man-machine, information, procedurallanguage, etc. ' Large manufacturingsystem vendors have also proposedCIM- architectures(IBM, 1987) as frameworksto developcomputer based manufacturing.
The scopeof the proposedarchitectures were limited, in the caseof ISO to 'discrete parts manufacture'. The ESPRIT CIM project was specifically aimed at mechanical engineering and machining operations 'because there are more manufacturing organisationswithin Europe involved in machiningoperations than any other single type of manufacturingand machiningrepresents the largest market for CIM system vendors' (Yeomans, 1987b). GRAI is restricted to a 'production managementsystem' according to Doumeingts and Breuil (1987). A commonthread to th architecturesis that they use ,e graphicalmodels to representthe various functions or aspectsof manufacturing.
Subsequentlymany 'architectures' have been proposed (Jorysz and Vernadat 1990a, Klittich 1990, Davis and Jones 1989, Graefe and Thomson 1989, Scheer 1992, Weston 1995, Los et al., 1992) for manufacturing enterprise applications. There is little
75 consistency in the approaches employed and the concepts used and the distinction between architectures, reference models, methodologiesand modelling frameworks is blurred. It can be seen that as architectureshave developed they have become more complicated in attempting to represent the breadth of manufacturing enterprise complexity.
5.2.1 CIM-OSA
The ESPRIT CIM Open SystemArchitecture (CIM-OSA) is an exampleof a complex all embracingapproach to manufacturingsystem architecture. The CIM-OSA is 'an open systemsarchitecture that defines an integrated methodology to support all phasesof a CIM system life cycle from requirements specification, through system design, implementation, operation and maintenance. Using a set of modelling methods a manufacturing enterprise can create a precise model of its own CIM requirements' (Jorysz and Vernadat, 1990a). The CIM-OSA architectural framework (or cube) is shown in figure 5.1. The diagram shows the CIM-OSA model generationprocess in the X axis of the cube, starting with a generic model using previously defined models (partial views) through to particular models of the subject under investigation. The model derivation processin the vertical axis of the cubeuses a conventionalstructured approach for software development. From requirementsdefinition through design specification to implementation description. The remaining axis of the cube bounds total enterprise modelling through four views; function, information, resource and organisation. CIM- OSA provides modelling constructs 'that take into account well-accepted ideas and principles such as SADT, IDEF and Chen's entity-relationship model' (Jorysz and Vernadat, 1990a). The focus of CIM-OSA is to use modelling methods to iteratively progressthrough each cell in the cube. To provide implementationmodels of function, information, resource and organisation,each processableby an integrating infrastructure (Jorysz and Vernadat, 1990b) that can be used to control the operation of a CIM system.
76 S7EPWISEhYSTAN714TION OFBUADWGBLOCKS
S7EPWISECM OFVEM
#4
STEPWISE DERIVATION OFMODELS
Figure 5.1 The CIM-OSA architecturalframework (adaptedfrom Jorysz and Vemadat, 1990a)
Two features are evident in the context of this research:
* That the complexity of the architecture and the scale of a complete modelling application (a simplistic view is that 36 integrated modelling initiatives are necessaryto populateall the cells) reflectsthe exhaustiverequirements of software or IS systemdesign for an enterprisesystem life-cycle. CIM-OSA is seenby its creators as 'a formal reference base available in the public domain' and that
77 'further exploitation must now be conducted under private initiatives' (Klittich, 1995).
Despitethe developmentof softwareengineering tools such as thoseproposed by Aguiare and Weston (1994) the completearchitecture and methodology,in its present stageof development,is of limited use for routine manufacturing enterprise modelling.
The notable difference between the earlier ESPRIT 'manufacturing architecture' approach and the CIM-OSA approach is the shift from a reference model of manufacturing defining functional organisation and interfaces to a modelling architecture and incorpomted methodology applicable in any manufacturing sphere with the constraints related to analysis of an enterprise being only function, information, resourceand organisationviews. The generic concept of the framework has the disadvantagethat its use in routine manufacturingapplications is more obscure than the referencemodel approach. It requires a depth of understandingon both the purpose of the architecture and of the range of modelling methodsused and in practical terms is likely to demandsignificant staff and computing resources. The referencemodel is more immediatelyaccessible and arguablydoes not require the sameintellectual skills to apply, however many referencemodels are likely to be required to provide a sufficiently broad referencebase for manufacturing.
5.2.2 The Zachman framework
The Information Systemframework was proposed by Zachman.(1987), who developed his 'methods and tools' framework for the information systemscommunity. It can be used to define the role of e)dsting and future modelling methods in terms of overlaps, gaps and the potential for integration. Zachman'soriginal framework proposal is shown in figure 5.2. It is intended as a set of information architectures rather than an architecture in itself Each cell in the framework is an 'architectural representation'that can be provided by one modelling method or an integrated group of modelling tools or
78 methods. The framework shown in figure 5.2 is specifically focused on software and systemdesign, it has been adoptedby the EDEFUser Group and developedto becomea framework for enterprise system design (shown in figure 5.3), still with a strong IS influence but using 'types of description' in six columns that identify modelling approachesto addressthe interrogatives;what, how, where, who, when and why.
Despite the different conceptsbehind CIM-OSA and the Zachman framework there are some common strands. The interrogative axis of the framework (in figure 5.3) defining What, How, Where, etc. can be comparedto the 'Z' axis of the CIM-OSA cube (in figure 5.1) that discriminates between Function, Information, Resource and Organisation. Similarly the perspectiveaxis of the framework can be comparedto the 'Y' wds of the cube discriminatingbetween Requirements, Design and Implementation.
The fundamental dfference between the two approaches is that CIM-OSA is a prescriptive software life-cycle methodology i. e. an architecture for system development in itself whereas the Zachman framework is a reference, aid or guide to IS systems inking.
The frameworkapproach offers considerable advantages for this research,it can provide a contextor structureto relateand position various modelling methods. It offerssome of the generality of architecturessuch as CIM-OSA but without intending to be a methodologyin itself A singleframework is not sufficientto provide a contextfor all modellingsituations, so like a referencemodel, a rangeof frameworks,albeit limited in number,could be necessaryto provide a comprehensivebase for modelling method applications.
Some proposals have been made to extend Zachmads original IS framework concept. Bruce (1992) suggestsa framework for manual systemsand points out that 'extending the (framework) concepts to consider other types of systems other than information systems is useful'. The IDEF User Group (1992) suggest 'non-information' system frameworks such as 'material systems'as potential extensionsand O'Sullivan and Brown
79 (1993) have proposed a modified framework as the basis of a classification of system designmethodologies.
Different Different types of descriptions perspectives 40
DATA PROCESS NETWORK DESCRIPTION DESCRIPTION DESCRIPTION SCOPE List of Entities List of processesthe List of locationsin DESCRIPTION importantto the businessperforms whichthe business business. operates (Ballpark view) MODEL OF THE Entity/Relationship Functionalflow Logisticsnetwork BUSINESS diagram diagram (Owners view) MODEL OFTHE Datamodel Dataflow diagram Distributedsystem INFORMATION e.g. entity/data architecture Nodefunction SYSTEM relationship e.g. (processor,storage, (Designers view) etc.) Linecharacteristics TECHNOLOGY Datadesign Structurechart Systemarchitecture MODEL e.g. Segment/row e.g. Computer e.g. Node (Builders Pointer/key function,screen or hardware/System view) deviceformats software,Line specification DETAILED Database description Program Networkarchitecture DESCRIPTION e.g. Fieldsladdresses e.g. Language e.g. addresses,link statements protocols ACTUAL SYSTEM Data Function Communicatins ,.
Figure 5.2 The Zachmanframework (Adaptedfrom Zachman 1997)
To advancethe understandingand use of modeffingmethods the frameworkproposed in section 5.4 provides the context for the applicationof modeffingmethods in manufacturing enterprises. A framework specificallyfocused on manufacturingenterprise modelling does not requirethe userto considerarchitectural concepts and has the potentialto simplifythe selection and routine application of modelling methods. It provides a pragmatic compromise between the complexity of architectural approachesand the resource and investmentnecessary to developa broad rangeof referencemodels.
so r. 0 ce Ln -- to) CU 99 0gn 0 Ei 8 r. 10 Z Z Ici Ici . o CL) txo rA .0 . - 9 4 -e I ci I ri l öt>o 4g ul 10(L) 2 0 2 tb *= bh -g 9 th4i cn
r. o '5 15 9a5 00 v9 gZ 1-1 Q M 10 ýe 181t4 CL) 0 rA Zi 0. ,. ä 0= .I tg txb 2 .Z200me 1 c 0. u --5 ', 0
Z m U Q= 8 .2 *ü A -; 0 0- 0 -8 (..), 0 m 0 (L)cu 0 1 0 *e '! ill, :N ýe i0 0 1 je *I 14lib;2ýcýii 40 0 es =2 0 :>A 1 .-0 .- cs ti
Ici
10.1 u 2 0 4) ..1 eä -U5 Z 1
0 (L) 9 § ck ;a (D= Cli @ - 0 "k5 404 R' 4. t 0 0 La
ce 16. z9 p4 Im cn
FA
2-, g
Q C)ýi vi iz
(D IZ, cl Z C) u2
81 5.3 REFERENCE MODELS FOR MANUFACTURING SYSTEMS
Central to the early phase of this researchwas the contention that referencemodels could provide a routine tool, to managechanges to organisationalstructure, to provide a meansto implement changesin manufacturingsystems or to designnew systems. In order to test the use and suitability of such an approacha reference(generic) model of processplanning was used to carry out a structured analysisand review of the processplanning activities of white goods manufacturer'. The rationale for the case study was an investigation of current practice to support the companieslong term intention to implement Computer Aided Process Planning (CAPP). As a precursorto the researchreported in this thesis a generic model of process planning (Colquhoun 1988, Colquhoun et al. 1989) for batch manufacture was developedto evaluatethe functionality of CAPP systemsin a CIM environment. The generic model has been used in this researchto test the use of referencemodels for manufacturing systems analysis and design and extends the work published by Baines and Colquhoun (1991).
5.3.1 The application of a EDEForeference model of process planning
This section demonstratesthe use of a reference model as a framework for capturing knowledge about processplanning in a batch manufacturingenvironment. (A brief review of processplanning for batch manufactureis provided in appendix9).
For this case study it was important to select an area of manufacturingthat containedmany of the featuresthat characteriseUK manufacturing. Processplanning for batch, piece part manufacture was selected as an ideal area to test the use of reference models, it has a significant human element, relies heavily on information establishedworldng practices and the experienceof the staff involved. In addition the case study company faces the sort of problems that exist in many areas of manufacturing; pressuresto reduce staff costs, new
1Cannon Industries, Wolverhampton.
82 technologies,an increasedneed for accurateinformation control of quality and the need to changeestablished working practices.
A generic IDEFo model is one that representsthe generally accepted or intrinsic activities that exist within a defined process or system and will represent all occurrencesof those activities. In this demonstrationthe generic model is used as a reference structure from which an AS IS model is developed. An AS IS DDEFomodel describesa situation as it exists in practice, that is, the inter-relationshipbetween activities, information, objects, people and facilities that make up a system,defined in company specific terminology. The AS IS and generic models provide a meansof examiningthe relationshipbetween activities to evaluate how a changein one activity may impact on other activities to influence the performanceof the overall system. An analystwith a previously preparedgeneric model of a situation uses the generic activities and ICOMs to structure the knowledge gathering necessaryto build an AS IS model. The AS IS model is thus developedusing company specific descriptionsof activities and ICOMs. The AS IS and generic models can then provide the basis for developmentof a model of the target system,i. e. a TO BE model using company specific terminology.
Figure 5.4 is the node index illustrating the extent of decomposition used in the generic model of 'Plan Manufacturing' where six levels were considerednecessary to clearly expose selectedaspects of processplanning. To construct an AS IS model using the generic model the context diagramis the starting point, in this caseA-0 - 'Plan Manufacturing' (figure 5.7, p.95). This would then be usedto identify the person(expert or 'mechanism' in IDEF terms) responsiblefor the activity. In the case where the mechanismis software then the person responsible for its operation is identified. The expert in turn identifies those experts responsiblefor child activities, in this case activities Al, A2, and A3 (shown in figure 5.8) and in particular A2 'Plan how to manufacture'. The activities Al 'Design product' and A3 'Plan when to manufacture',although an essentialpart of the context, are outside the scope of processplanning and remainas the context of the AS IS model.
83 A-1 Producenew product A22241 Selectmachine A-1 I Market and manageproduct A22242 Derive processsequence options A-12 Plan manufacturing A22243 Derive capacityrequirements A43 Executemanufacturing program A22244 Assignsequence priority
A-0 Plan manufacturing A223 Establishauxiliary requirements
AO Plan manufacturitig A2231 Selecttools A2232 Retrievetool information Al Designproduct A2234 Establishworkholding requirements A2235 Retrieveworkholding information All Establishdesign concept A224 Establishoperation information AIII Producedesign ideas Al 12 Selectdesign principle A2241 Establishoperation details Al 13 Evaluatedesign Al. 14 Formalisedesign A22411 Retrieveoperation data A22412 Analyse operation A12 Carry out functional design A224121 Establishcomponent criteria A121 Identify major assemblies A224122 Derive machiningparameters A122 Retrieveassembly design A224123 Establishoperation characteristics A123 Establishcritical designfeatures A124 Designassemblies A22413 Updateoperation data A125 Finaliseproduct design A2242 Generateprograms A13 Carry out detail design A22421 Establishcutter path requirements A131 Identify componentform A22422 Producemachine movement data A132 Retrieveexisting/similar A22423 Producepart programs A133 Design component A224231 Establishsystem requirements A2 Plan how to manufacture A224232 Postprocess A224234 Provepart program A21 Plan product assemblymethods A2243 Derive operationtunes A211 Analyseproduct - A212 Establishassembly technique A22431 Derive machiningtimes A213 Establishassembly requirements A22432 Retrievesynthetic times A22433 Produceoperation times A22 Plan componentmanufacture A2244 Format information A221 Analyse component A3 Plan whento manufacture
A31 Produceaggregate production plan
A311 Derive productionlevels A312 Assessresource capability A313 Establishresource requirement A2211 Retrievecomponent analysis A314 Establishproduction plan A2212 Separatefeatures A2213 Derive featuredependant geometry A32 Establishmaster production schedule A2214 Determinegeometric component elements A33 Establishmanufacturing and resourceplan
A222 Derive manufacturingmethod A331 Establishcomponent demand A332 Establishpurchase requirement A2221 Determineraw material A333 Derive manufacturingrequirement A2222 Selectprocess options A334 As capacity A2223 Selectprocess A335 Establishmanufacturing program A2224 Selectprocess sequences A34 Scheduleresource and manufacturing
Figure 5.4 IDEFONode index for Plan Manufacturing (source:Colquhoun 1988)
84 The expertsresponsible for A2 activities (shown in figure 5.9) identifies those responsiblefor A21 'Plan product assemblymethods' and A22 'Plan component manufacturing' activities. Experts associatedwith A221, A222, A223 and A224 (shown in figure 5.10) are next identified. The generic model is focused on processplanning for manufacturedcomponents and thus the decomposition of activity A21 'Plan product assembly methods' is not considered. Similarly all expertswho can contribute to defining processplanning throughout the decompositionare identified and subsequentlyinterviewed. The interviewing processcan either take place immediately an expert has been identified or after all experts likely to contribute havebeen found. Naturally the latter hasthe advantagethat conflicts and overlaps can be dealt with before the interview process starts. In practice the expert is questioned using appropriateactivity diagrams. The objective of the questioningis primarily to make the activities, controls, inputs and outputs of the generic model company specific. This means finding preciselythe documentor terms usedby the company. At the sametime additional or missing inputs, outputs, controls and activities may be identified.
During the interview processthe expert is shown the appropriate activity diagram and is prompted by the int erviewer for companyspecific information on an activity by activity basis. This procedure is completed for all the activities within the generic model. In simpler processplanning situationsthe number of activities could be less than the generic model but should not normally be more. As activities and activity diagrams changethrough different levels so may the experts change. Indeed it is critical that experts involved with eachactivity are those interviewed so that the true AS IS model can be identified.
5.3.2 Deriving an AS-IS model using a generic model of process planning
By definition the genericmodel of processplanning should be able to cope with any type of manufacturingsituation. This proved to be true using the model to derive an AS IS model in a case study company. The companyproduces domestic cookers, employs 800 people and has a turnover of in excess of Orn (in 1990), with a process planning f1mction that concentratesmainly on press work for sheet-metalcomponents. In all, five experts were identified and subsequentlyinterviewed. The study took a total time of 5 hours. It is
85 significant that the processplanning functions were completely recorded by the researcher with no previous experienceof the company and in collaboration with experts having no previous experienceof the IDEFo method.
Duringthe interviewingexercise thirteen activity diagrams from the genericmodel were used comprisinga total of forty-sevenseparate activities. The samethirteen corresponding companyspecific activity diagrams,this time with only forty-four activities,were necessary to fully describeprocess planning in the companyand provide the AS IS model. The same nodetitles were usedand in practicethe genericmodel diagrams were -simply annotated to indicatethe companyspecific terms and differences from the genericmodel.
5.3.3 Conclusions on the use of a reference model
The purpose of the case study was to investigate current process planning activities in the context of the introduction of a CAPP systemand to use the study as a meansof examining the use of a referencemodel for a manufacturingsystem. Figures 5.7 to 5.13 are used to illustrate the important findings from the casestudy. Node numbersfor Generic and AS IS model diagramsare suffixed A and C respectively.
Using the genericmodel proved invaluablein identifying those experts at various levels in the organisationthat were able to assistthe interviewer in the data gathering process.The three experts responsiblefor the generic model activity A22 'Plan component manufacturing (in figure 5.10) are identified in the correspondingAS IS diagram,the Work Study manager,the Production engineeringmanager and a Project and CAM systemsengineer (shown in figure 5.11, node C22 'Plan componentmanufacturing'). The subsequentdecomposition of node A222 (figure 5.12) to A2221 to A2224 was usedto developthe correspondingnode C222 to activities C2221 to C2224 (shown in figure 5.13). It showed that the same experts were directly responsiblefor lower level activities such as C2221 'Determine raw material'. In the lowest levels of the AS IS model some devolution of responsibilitytook place. For example the company relied on a machine setter as the expert responsible for activity C224122 'Derive machiningparameters'.
86 Using the individual activity diagramsin the interviews was found to be an excellentway of stimulating the expert into providing a thorough review of activities. The diagramitself also structured the recording, critique and correction process directly using the box and arrow notation. It is possiblethat using the diagramscould constrainthe replies the expert gives so it is important to emphasisethat the diagramsare a framework for discussion.
Throughout the generic model specific company terms were identified as replacementsfor the generic terms where possible(shown in bold italic text in AS IS model diagrams). For examplein the AS IS model activity diagramC22 (figure 5.11) activity C222 is subjectto the control 'Design changenote' and activity C221 has the control 'detail design drawings and parts list' whereas in the equivalent generic model activity shown in figure 5.10, these controls were describedas 'Method changerequest' and 'Manufactured componentdesign'. Similarly in C22 (figure 5.11), activit-y C224 has an output 'Master op. card' that has replacedthe 'Operation information' usedin the genericmodel activity A224, in figure 5.10.
The AS IS model exposedextensive use of heuristics for example, activity C224 'Establish operation information', (figure 5.11) has 'Process and material information (fleuristic)' as a control in contrast to the equivalent generic model activity A224 in figure 5.10 which has 'Process and material information' as a control. This drew attention to the practice that the staff carrying out the activity madeno referenceto standardsor companydata, but relied on their unrecorded experienceor estimatesto make decisionson operation parameters. Too many heuristic controls on activities flags a potential vulnerability for. the company which could result in a dependencyon individual staff and potential implementationproblems if the company decidedto use computeraided process planning in the future.
Nfissing controls or inputs suggestthat the system is operating in an open loop mode that could lead to uneconomicoperation. For example,in establishingthe AS IS activity diagram
C222 (figure 5.13) the information 'Aggregate annual demand (verbal)' was identified as a control on all activities using the equivalent generic model information 'Component quantities and problems' shown in figure 5.12. This comparisonexposed that the source of the generic model information was a result of activity A3 'Plan when to manufacture' (figure
97 5.8) based on actual batch sizeswhereas the company relied solely on aggregatequantities basedon marketing estimatesof annualproduct demand.
The generic model also highlighted the company's use of process planning activities to produce outputs not normally associatedwith processplanning. For examplein figure 5.13, activity C2224 has an output 'Capital expenditure application' and activity C2223 has a control 'Product tool budget', no direct equivalent control or output exists in the generic model diagram in figure 5.12. The AS IS activity involved financial evaluation of capital plant and resultedin a formal documentedapplication for capital expenditure. A processnot consideredpart of the routine processplanning activity when building the generic model. A responseto this could be to modify the genericmodel for future use.
A comparisonof the numberof activitiesin the Genericand AS IS modelrevealed that the AS IS modelcontains 3 feweractivities. For examplein the genericmodel A22411 'Retrieve operationdata' (seethe nodeindex of figure5.4) hadno equivalentin the AS IS model. The reasonbeing that the companyre-defined its operationdata each time a new componentwas planned. For the samereason activity - A22413 'Update Operationdata' also had no equivalent. Similarlythe diagramA22423 'Producepart programs'in the genericmodel containsactivity A22423 I 'Establishsystem requirements' before 'Post processing',since the companyhad only one programmablemachine and thus a singlesystem the activity or its equivalentdid not appearin the AS IS model.
Previous knowledge of the IDEFO method was not found to be necessary for those interviewed. Nor was there a need for researchersto know the company organisation or operating methodsbefore undertakingthe investigation. The approachproved to be good at drawing attention to weak practicesin the processplanning functions acrossthe organisation. For example,it becamequickly obvious that processplanning at the companyhad only weak links with the Production Control departmenthighlighted by the use of aggregatedemand quantitiesreferred to earlier. In addition the most recent processplanning activities relating to CNC programmingwere developingentirely independentlyof the manual systembecause the Project and CAM system engineercarried out all planning related to CNC machining.
88 The AS IS model also showed that staff with the title manager were also found to be working very much at an operationallevel in processplanning activities.
Of particular concern in generic modelling is the depth and breadth of decomposition necessary to provide sufficient detail. The node index, figure 5.4 shows that the decomposition of nodes Al and A3 provided sufficient detail after decompositionof three levels. Whereasto describeprocess planning activities fully, node A2 has been decomposed to six levels. The breadth of the model was limited by the need to identify those interfaces necessaryto support processplanning activities. On the other hand the depth was limited to retain generic descriptionsof activities. It was consideredthat taldng the model to finther depth would introduce the needfor non- generic activity descriptions. This is a limitation to the use of genericmodels as referencemodels for low level detail.
The diagrams provide structure to interviewing and, using a valid generic model, are exhaustivein terms of identifying information flow. It could also be claimed that less skill in IIDEFomodelling is necessarybut practice is still necessaryto conceivevalid generic models. The generic model also has the potential to limit the variability of different analystsif they were to addressand model the sameproblem.
This application of the generic model servesas an example of the capability of 11DEFoas a general purpose functional modelling technique to provide a clear picture of a complex aspect of a manufacturing organisation. The model could be developed to interface with other models of other areasof a manufacturingorganisation such as product managementor manufacturingto provide a complete generic model of batch manufacturing. The limitation of this approach is the potential number of models that would be necessaryto cover all manufacturingsystems. However somerecent researchhas pursuedthe use of 11DEFomodels as referencemodels for manufacturing, Maull et al. (1995) for example, his EPSRC Grant GR/J 95010 'Good practices in Business Process Re-engineering in manufacturing engineering' has DDEFomodels describing standard business processes as a deliverable. The US National Institute for Standards(NIST) proposestemplate-driven systemsdevelopment with 11DEFoand IDEFx models using 'a Template Re-use Library containing (IDEFO and for IDEF, x) models which have been developed,tested and approved as standardtemplates
89 use in particular classesof systems,or systemtypes e.g. payroll system,radar system etc.' (Law, 1992). Interestingly the researchalso proposesthat 'best of breed' models could be reverseengineered to produce standardgeneric modelsfor re-use.
5.4 A MANUFACTURING ENTERPRISE MODELLING FRAMEWORK.
This part of the research presents a proposal for 'a manufacturing enterprise modelling framework to provide a structurefor managingchange in a manufacturingenterprise.
To selectand use a modellingmethod it is importantto establishthe capabilityof the method in the contextof alternativemethods and in termsof thoseaspects of an enterpriseit hasto describe.To justify new methods,to proposedevelopments to existingapproaches, and to integrateexisting methods demands a referenceframework from which modellingcapability can be evaluatedin a manufacturingcontext. The framework is a graphicalway of presentingand organising modelling methods having different notations and syntax. It differs from the CIM-OSA architectureapproach that proposesa 'finite but comprehensiveset of modellingconcepts' (Jorysz and Vernadat, 1990a)to provide an integratedproce=ble modelof a manufacturingenterprise. The frameworkdefines a viewpointfor interpretingthe role of a modellingconcept and as a basisfor the classificationof modellingconcepts.
The rationalebehind the manufacturingenterprise modelling framework proposed in this research(shown in figure 5.5) is the need to develop a framework for manufacturing enterprisesthat does not have the strong IS characteristicsof existing approaches. It providesa contextfor the use and developmentof modellingmethods where the object of modellingis to achievemanufacturing system change and improvementwhich may not be specificallyrelated to softwaresystem implementation or design. Threemodellingfocuses are representedas columnsin conjunctionwith four rows representingperspectives of an enterprise. Thefocusesindicate the key aspectsof an enterprisethat could be modelled. The intersectionof columnsand rows givesthe twelveframework cells numbered from 10to 120. Each cell distinguishesan aspect of a manufacturingsystem from a particular
90 perspective andfocus. Modelling methodsor tools are mappedon to the appropriate cells, there is no significancein the ordering of cells and no methodology is implied by the cell numbers.
Column a: Models of 'configuration' answer interrogatives such as what?, who?, where?, i. e. the objects, material, products, information, data and resourcesused by an enterprise together with their relationships. For example,what resourcesare needed,who uses them, and what information and materialsare necessaryto run them.
Columnb: 'Behavioural'models describe how manufacturingis conducted,for exampleto representthe manner in which the enterprise operates, how a product is produced, when eventsoccur, what initiates,enables, constrains and inhibits manufacture.
Columnc: 'Quantitative'models predict measuresof enterpriseperformance for example the quantityof productsproduced for a particularcapacity, what the manufacturingtimes are, how many staff are necessary,volumes of material flow and other measurable performanceparameters.
The four framework perspectives are approaches an analyst can take in modelling a
manufacturingenterprise:
Row 1: An 'enterprise' perspectiveis a high level view that considersthe manufacturing
scope of an enterprise, the products, its overall goals, operation and values and the
manufacturingstrategy that it usesto achieveits goals.
Row 2: A 'structural' perspectiveexamines the relationshipsbetween the elementsof an
enterprise. It is a static functional view that describesthe interfaces,links and relationships
that exist. For examplea structural perspectivemodelling approachto the configuratiOn Of
91 a manufacturingprocess indicated in cell 20, might define the information required to run the facility suchas a works order, a computerprogram, a processsequence together with the
FOCUS a b c CONITGURATION BE11A"OURAL QUANTITATIVE What,Where, Who How,When HowMuch, How PERSPECT1W IEWF"RISE 10 Values, 50 90 GoaL%Scope
2STRUCTURAL Functional 20 60 100 Relationships, Interfaces 3 PROCESS 30 70 110 Thestages involved in informationand materialflow 4DIWAAHC 40 80 120 Ileactivation of informationor object now I ENTFJZPR1SE Product,Material, Constraints,Material Resourceconsumption, OPERATION Resources,Locations, flow, Schedulesand Output,Performance InformatiorL thing.
Figure 5.5 The manufacturingenterprise modelling framework.
material and personnel necessary. A structural perspective modelling approach to the behaviour of the same situation, indicated in cell 60, would identify at what stage information was needed or whether it was required simultaneously, and who initiates, monitors, stops and maintainsthe process.
Row 3: A 'process' perspectivelooks at those aspectsof an enterpriseconcerned with the actions that take place to changethe state of material or information and the sequenceof events and the causeand effect of events. A process perspectiveof the coilifiguration of a manufacturing process indicated by cell 30 might, for example, describe the sequenceof
92 starting, loading and running and any alternative associated processing sequences. A behavioural focus and process perspective, cell 70, might examine the temporal aspects, tooling and information availability and decisions that constrain the various alternative processingsequences.
Row 4: A 'dynamic' perspectiveviews the representationof the actuation of aspectsof an
enterprise to expose material, information or object flow. A dynamic perspective and
configuration focus of a manufacturing process indicated by cell 40 could examine the
relative positions of material, processesand tooling by modelling material flow. A dynamic
perspective and hehavioural focus, cell 80, might model material flow to examine the
implications of variousscheduling approaches used by the process.
5.4.1 Conclusions from the manufacturing enterprise modelling framework
It could be argued that the framework, although another way of classifying modelling
methods, has little value other than intellectual neatness. However this researchproposes
developmentsto the IIDEF methodsthat neededa context for their developmentto outline
relatively the role and extent of modelling methods in manufacturing systemsterms and to
see the significanceof the modelling developments. The framework, through its two axis,
provides the fundamental'viewpoints' and 'purposes' of modelling methodsthemselves from
which evaluationsare possible. In an application context the framework is a managementor
modelling strategy that fies above the modelling process itself where the issue is the part
modelling plays in effecting a change. The framework can also be used as a common
representationwhere the modelling possibilitiesi. e. the various viewpoints and contexts can
be related to the objectives of problem solving (the researchhowever does not attempt to
prove this point). Its simplicity relative to the frameworks that have influenced its
development, Zachman and CIM-OSA, re-enforces a case for its routine use. The
93 framework could be developedin a t1ird axis using Generic, TO BE, and AS IS cells but
further partitioning addsto the complexity without contributing to understandingor clarity.
The intersection offocus andperspective axis definestwelve cells. Each can be assigned(or mapped) to a modelling method to characterisean understandingor carry out an analysis from a perspectiveand areaoffocus boundedby the particularcell. Cellscan be seenas a be - taxonomy of a manufacturing enterprise against which IDEF modelling methods can mapped as shown in figure 5.6. EDEF modelling methodsmay only partially populate cells and may overlap cell boundaries. IIDEFohas a significant capability in cell 20 and a partial capability in cell 10 and 30. The capability of IODEF3(Process flow networks) is largely confined to cell 70 with partial capability in cell 30. The ideal is to populate as many cells as possible with appropriate, single or combined modelling methods. The purpose of the modeHingapproach proposed by the research(reported in chapter 6) is to synthesiseIDEFo and IIDEF3 extending their scope in an integrated method that has the potential to significantly contribute to modelling in cells 10,20,30,50,60 and 70 of the framework.
FOCUS a b c CONITGUA47ION BEIL07OURAL QUANTITATIVE What,Where, Who How,When HowMuch, How PERSPEM Quickly,How M I ENTERFRISE 1 10 50 90 Values, XDZFo Goa][s,Scope
2 STRUCTURAL 20 60 100 Functional XDEFo RelationshiM Interfaces 3 PROCESS 30 70 110 The stagesinvolved in XDEFo TDE.F3 infonnation and 1DEF3 material flow . 4 DYNAAHC 40 so 120 The activation of infonnation or object flow
ENTERPRISE Product,Material, Constraints,Material Resourcecomurption. OPERATION Resources,Locations, flow, Schedulesand output, Pesformance Infonnation. timing.
Figure 5.6 The manufacturingenterprise modefling framework and IDEF methods.
94 s. -u-ar- cu 2(1) i pý 0' mE 9 CL cr m .2 i! E E rý um vi ab c % E --0 -0-5 Z
CL
2 Im 0.
F- :ý u ýd ýI M ý 0
,0- 25 Z 3: 0 20r_ Z-
-0 ýu .
Co j! 0
E 110 r- cec CL 10 . 9 x «a c; 0=m 0 M r- 0 E a- 0 mü- (L) (Doý 4) 0 > Ki .2 "
0 20c ý7 > 9 <
4i c?
Figure 5.7 Node A-0 Plan Manufacturing (the context)
95 0 c0 (D :, C» : " e .0 :5 r_ ;5-, ,- 0M u 2 m j2 m 2 E', 2 . M-Z to CLZ tr2 Ck. 'IM r- «a cm0Z E A Co 2 JM
0 Z X]
0 2 CL - lg m le w o cm 06 Q Z 0 r-
0 ýI 0 < u m Co mm3 Cd w ý:) (0 ý: im m X - 00 Cd 11
7 0m Cib
> 40
00 (P to (0 to A 2-Sz ED c0 Emr 0 Z CY0. ME to ob _ZE tu c E od 0 M. 0 E ci a .li el- - l> M - (D ge x 9 w 0 0 0: 0 c% 0 - gl. 00 m c
KL 's \o 90
r- (L) , cm -a C 113 PZ 0 2 r4 = . :Z E00 ,
M 0c 3: a IM _ CL = ý,' 2 cm c6 C» 3 00 Z 0 1 0 *0 0 r- 0
l z
Figure 5.8 Node AO Plan Manufacturing (the viewpoint).
96 0) ., 0 11 C r- 4) 1! r: 2w "-(3 CO)
0) ab
E
CLein rrk.-
Q Z Z 0,
c26 to ý (Lb . m E 10 93 Cd , :D m ýx 0.1 ,- ä , 0,
r- im Cb -0
me 0 15
c% 2 0-. 00 0 t- m 1 u 10 Uli >b wi
0 A. . 93 . r g2. U- ýA 0 93 ein 20 Z 0 r Co 0 T 0 0.4 e4 < A; - lu «0 Z 11 CL E 0 Z Z0
4) CL Ck. E
KL 0 re Z) 1 1CZ,
Figure 5.9 Node A2 Plan how to manufacture
97 ZZ
tM t3o 2 1.01lco H ID m Eo 16 3:. 0 0 ) M 06 tr 0) 5 2 r3 m .0 41 z tr CL Cý
Im .0 to - ;; 2 (DU) . cEii q2 c t3l 0 4) 0. I-L 2 li .-Z; _0 'd E m o. EG ID w tx CL -0 r O. S
00
Z P0
W E. 'a OU ý U tx, .1- ' 0 =0 U r- = c0 0 17 co =
I - - --* 10 W=c -3 cl -6 JC 00
2 tv be > C 4) 0= CL- C E 0. CX 0 u tr tx
bl) M P, 2 0 'a to W 92 0 sa. . 0- IL
1 00 (D CL r 7 co ýo 010 E V 0 o. E Im 0 ý; 2-0 10 -E c re 0 4) 0.2 0Cý o.. G r- > r -;; a0 0 a ID 06 M.0- 2 th CD 0 BE 0 0 W0 2 a: 0c _o z 93 0 - 0 :a 0 AS '00 11 4ý 19 E0c 41 5E 0 tm 0 7jj CL rA _j E" 00 co to 0 0 l zI
Figure 5.10 Node A22 Plan componentmanufacturing.
98 E c
's r 0. KT m CL Zr 2 (L) M 2 LU
< Co
m 42 ri WO > gý :%Jä- .2 c3 4 -e '0"" 2 Co 4-, x
gz Z Z 0 < C. 2 r 1 0 E 0 :2 0 od w Z) 0 e7 O= 0 dý- t; ý: . :@g 1 Im, od 2e0 0, -a d0 x 5e 14 b . 'oýto 0) - 0ý0.0 e r > m 1 t, Ct
;e tu 1 1 JE 2 4i Cb lu r- -Itz ýo 0 '% 0 .. - - - - (00 In > ,0 RL A !4 E 0 - 0 u " 0.2 -Zt) 0 0 42 4. 0 :3 - (6) e
.2 0. cm 0 0 1 c " Jg E 5 o. E to 0 L C*4 00to en I re rn IM KL t: Z E0c '» 0 0E 0 0 00 Cd Z 'Z. tu .--Z- --0 .2 i -äi cl c:w -2 _Z < 13 0, 0 t, e , -OM tN e _j 0 .2. cz zz woJE 46. w a Jit0,. 0 rA 0 10 2 ri im c3 CL0 0;i 1 ID Z z - -
Figure 5.11 Node C22 Plan componentmanufacturing.
99 a Cb 1Z5 3 cto r- 0
1 U A to cb tu 0) 0. cm '. A c4 11 lo 0,5 * ji
M 020
1,3 27 9 In
< 0 gx w Z)
x 0.0 0 je
0 r4
MEr u (L) ýJc 0 3M0, -
0c- ; ý a. 0 -0 - .-
0- (1)
122 9 en (P --
M ýz 0 E -- ei
p4 EE
_M 4)
%M CL 0 926 :sE 0 LZ4i ME (DO- In 0
Figure 5.12 Node A222 Derive manufacturingmethod
100 00 a)0 ý c A: xZ
'a qm E c E L) (0 il-: 3e c to00 3,2 0.- L L» (L) Z clý Qý *oll 2 ÜA .0 220,5 w 0.0 0 Z a- 80 F- 0 2 3 n. 0. :2 < c c im 0 Co15
w Z) im e F < IA er- ý
ýo .
r
Z, t rn
a U.C LU < UM0: U % im r4 9 Zf e 0 i 2- 0 a 10 to& 0-0. E =
g', 0a LU 4)
rE *.. " --101
44 t-,E Q mo 10 ü; - - 0 m - c2w. r. 0Um ;i Ke''ýýr A g: ' C - x tý 00 0 ý- 0 F-w EE D 0 00 < Cd Z
p 3 (x. -1: * 1 tu 0 Q 14-5 ;5E u Z 9
0. 121 P) 1 -
Figure 5.13 Node C222 Derive manufacturingmethod
101 CHAPTER 6
A COMPOSITE BEHAVIOURAL MODELLING APPROACH
6.1 INTRODUCTION
In this chapter a synthesis of the establishedIDEFo method and the IDEF3 process description method is describedin the context of the manufacturing enterprise modelling framework. The composite approach,-called IIDEFO-3 (Colquhoun and Baines, 1994, Colquhoun et al., 1996), is proposed to increasethe acceptability and promote the use of modelling in manufacturingenterprises by extendingthe scope of individual 11DEFmethods to provide a behaviouralview of manufacturingsystems that can contribute to modelling in cells 10,20,30,50,60 and 70 of the manufacturingenterprise modelling framework. The combination of methodsis usedto provide the additional capability to model concurrent and simultaneousactivities, the nature of constraint dependencyand the temporal aspects of manufacturingsystems.
Three stagesin the developmentof the IIDEFO-3method are described:
Stage one is an account of a case study that demonstratesthe potential for the modelling approachthat uses IIDEFodiagrams to configure an IDEF3 analysis. The two methods are shown to be complementaryan& the case study provides an example of how the two methodscan be usedto structure and potentially speedup the analysisprocess.
102 Stage two exploresa formal approachfor integratingthe two methodsand proposesa Meta- model to show where links betweenthe graphical elementsof the two methodsare necessary and to re-enforcethe casefor integration.
Stage three usesthe findings from stagesone and two in a casestudy that demonstratesthe applicationof the conceptsbehind IIDEFO-3.
The chapter concludes by presenting the IIDEFO-3 approach in the context of the manufacturingenterprise modelling framework prior to a casestudy in chapter 7 to validate the approach.
6.2 A BLUEPRINT FOR ENTERPRISE MODELLING METHODS
The investigative work of chapter 2 and the analysisprovided in chapters 3 and 4 has re- enforced the view that enterprise modelling approacheshave largely developed from IS methodologies and are in general focused on data and information flows and transformations. The exhaustivedetail required by IS approachesis not necessaryto the same extent in manufacturing enterprise modelling and IS approacheslack the facility to model important manufacturing system characteristics. Modelling behavioural aspects of manufacturingsystems is critical for guiding the changeprocess.
The manufacturingenterprise modelling framework described in chapterfive hasprovided a manufacturingcontext for 'behaviour. The behaviourcolumn (figures 5.5, and 5.6) is focusedon the mannerin whichan enterpriseoperates. It considers;decisions, the causeand effectof alternativesequences of events,what initiates,monitors, stops, enables and inhibits. Timingis an importantaspect of behaviourand the modellingconsidered in this researchhas beendefined as staticgraphical modelling (section 1.3) which precludesthe facility to model changesover time. However a static approachcan provide a temporaldescription that placesevents, processes, functions, etc. in relativechronological order. Sucha description canbe a precursorto dynamicmodelling.
103 Manufacturing enterprisemodelling shouldnot be seenas an isolated activity carried out as a fragmentedconsultative exercisebut as a routine elementof ýhemanagement process. For a modelling method to contribute and gain acceptanceas a routine part of planning and managementcertain featuresare essential.
I It must be able to explicitly describe an aspect or aspects of an enterprise in a way suitable for managing change in manufacturing where a knowledge of functional structure and process behaviour are necessaryprior to implementing a change or enhancement.
2 It must be capable of capturing and representingthe perceptions of participants in a systemin a way that makesit possibleto identify qualitative or quantitative differences between those perceptions. This ability is necessaryif a method is to have sufficient resolution to allow meaningful analysisand to provide the potential for optimisation. The method should also be capableof aggregation and decomposition but should not encouragethe distortion of a systemdescription to fit the characteristicsof the method.
3 The resultingmodel or diagramsshould be simpleenough to be used as a basisof communication between system participants and the analyst. The method needs in its initial form to be graphical, capableof being producedwith or without a CASE tool and having a minimal number of diagramsto complete a model. For intuitive use simple graphical constructs arerequired negating the need for an extensive knowledge of the modelling technique.
4 The modelling method should require minimal time to gather the information necessary to constructdiagrams, to build modelsand to validateand analyse them.
5 It should be useful in the changeprocess in allowing decisionmaking to take place using more complete and realistic information. The model must also facilitate analysis and evaluation to assistin predicting the perfonnanceand behaviour after a changehas been implemented.
104 6.3 STAGE 1-A CASE STUDY USING IDEFo AND IDEF3
The case study reports an investigation of a manufacturingcell engagedin manufacturing and pacIdng a fabric based health care product. The work was part of an initiative by the company' to simplify processesand organisationwith the ultimate objective of improving product quality. IDEF3 was used to structure information gatheredfrom participants in the cell and to use the resulting description to assessthe impact of machine problems and contribute to improving cell performance.
The work of gathering and validating the information took a total of eight hours and was done by the author who had no previous experienceof the company. Information gathering focused on how the processesof the cell were actually carried out using an approachloosely basedon that of Mayer (1993b) i. e..a knowledge captureapproach to identify the following:
" The individualprocesses involved. " The conditionsto start,maintain and stop each process. " Responsibilityfor the process. " Precedingand subsequent processes and the logic of processes. " The natureof the process,i. e. timeto complete,or type of occurrence; continuous,repeating, single instance. " The information, verbal or documented,required to carry out the process. " The objects that take part in the process,tools, material, resources,etc. " The constraintson the use or involvementof objects. " The exceptionaloperating conditions of a process.
The initial investigative work using the 11DEF3method exposed a 'pool' of twenty-five related processesand it was evident at that stagethat a structure for building PFN diagrams was necessaryto resolve conflicting narratives from the staff interviewed. An IIDEFo diagram was therefore producedto quickly provide a functional view of the systemwithout the constraint of consideringprocess logic.
1S. choH ConsumerProducts, Derby
105 The IDEFOmodel of the cell (called ZENO) was produced to provide the framework. The initial diagram in the IDEFo model is shown in figure 6.1 node A-0 'Run ZENO', this shows purpose, context and viewpoint of the IDEFo model. Decomposition to AO, the viewpoint identified diagram, six aggregatedactivities Al - A6. The diagram is shown in figure 6.2 describing relationships between the activities in terms of objects and information. The diagram does not attempt to indicate the logic of activities, for example, activities A3 'Run 400 press' and A4 'Monitor process' are carried out concurrently and both are initiated at ihe sametime. The IDEFo diagramcannot explicitly depict this aspectof cell operation.
It does however provide a framework for building an IDEM model. The processes establishedin the IDEF3 model building exercisecould now be assignedto activities in the IIDEFodiagram and an equivalent'Run ZENO' IIDEF3 (PFN) diagramwas produced (shown in figure 6.3).
Proceduresand standards
Waste material and product Matedal and Consurnables Reports Packed product
Purpose: To Improve product quality through simplification of processes and organisation.
Viewpoint Cell management and organisation
Figure 6.1 Node A-0 Run 2ENO
106 0 rL «a cu 0he :1 0-, (0 -0 -ti t IL 0.
CL
CL
4 10
c
4, N
vc ýo IOD
-6a '. t CL it in CIO W5PC c0 .0
0 CL CL x CL 4)72 75 16- tu Cu 12 «a g LL lu Z CL sw (D CL 11416 0.2 -E!-at! 0 8 .2
0 ca :3 :3 -t; (a M R E 3: a a. Z) B 00 20 . CL-L CL- CD A 13 x v 120 rL. tf
(4) CL ofN .0
CL 0M -0 c
od to
C; ILP-
Iti 12
CL
tn 0 0: 5 E
0 Figure 6.2 Node AO 'Run ZENO'
107 The IDEF3 method is a process description capture method where process information, process logic and sequencesare gathered piecemealfrom system participants to build a larger picture, the method is not designedto model conceptualactivities. It is a bottom up approach in contrast to the top down approach of IDEFo where aggregated, (often conceptual)activities are a model starting point.
obtldl Sd 400 1 LL4 jj! 'It. k-!: up LA j . -T-4 RIM400pno W= 0 Fý -ý ,. id T -..ij 1-24
4L2 mibwPOd &I Lfi PackprodW - 4. "
U Monk= V proý
Figure 6.3 The IIDEF3PFN 'Run 2ENO'
Each activity (orprocess) in the IDEFo diagramformed the basis of discussionwith the cell expertsbased on; firstly agreeingon the aggregatedactivity description and the objects and information used in the activity. Secondlyby grouping activities from the processpool into aggregatedUOBs.
The EDEF3 description of the cell describesthe logical and temporal relationships of the In figure 6.3 aggregatedprocesses. the processes,2 1.1-'Obtain material' and 1.1- 'Set up 400 press' are sequentialand the occurrenceof 'Obtain material' must be complete before the 'Set up 400 press' process can take place. The asynchronousfan out junction J1 indicatesthat the processes'Run 400 pressand Return material', 'Inspect & pack product', and 'Monitor process' all take place concurrently but do not necessarilystart at the same time.
108 In total forty three processeswere identified and are depicted on six PFN diagrams (presentedin appendix7). The model was decomposedto three levels to exposesufficient detail (in figure 6.3 shadowedboxes indicate that a UOB has been decomposed). OSTN diagramswere not consideredin flis application.
An Elaboration Document was produced for the 'Run 400 press' process that was the aspectof the processof particularinterest to the companyin this application.
ELABORATION DOCUNIENT UOB Label: Run400 press UOB Number: 2.1 Objects: Platingpaper OperatorI ReferenceTile Operator2 Work to list (batchsize) RovingInspector Material Line supervisor- Solvent FinishedProduct Facts: Throughputfixed at 90 products min Fault freerun timesvariable and not documented Predictedefficiency 63% Faultstypes are not documented Operatorsinformally exchange tasks eve 2 hours Constraints: In-feedPlating paper available To initiate -GuardsClosed In-feedMaterial available Provingrun accepted To stop- Batchcomplete To confinueprocess- Shift end Productquality acceptable Supervisorstop (split batch) (Operatoror rovinginspector decision) Machinefault Unacceptable I product
Figure 6.4 A UOB Elaboration documentfor'Run 400 press'
It was evident when gathering the information for Elaboration Documents that there is a closecorrelation between it andthe informationnecessary to build an ICOM structurefor an IDEFo diagram. This fact was not pursuedin this casestudy but it is consideredin the proposalsdiscussed in section6.5.
109 6.3.1 A review of stage 1
Two aspectsof the casestudy can be examined(i) the behaviour of the cell and (H) the capability of a new approachthat correlatesIIDEFO and IIDEF3 models. (i) Three findings were revealedas a result of modelling; e The application of IDEF3 revealedthat the line operators were largely monitoring raw materialand finished product in orderto respondto frequentmachine operation faults. It exposedthat an operatoron the output sideof the pressis requiredto checkseven featuresof eachproduct and if necessaryreject it in a completely unrealistictime window of only 0.7 seconds. The operator rejected a batch of componentsby sweeping the output conveyor in the vicinity of the faulty items thus rejecting many acceptable components.
It exposed that the operator on the input -side of the press unnecessarilymonitors previously inspectedin-feed material.
Gatheringinformation and building the processdescription provided an objective,structured view from which processrationalisation can proceed. Those processesthat can be combined,that are a responseto machineperformance problems or that can be carriedout automaticallycan be examinedin moredetail. ThePFN diagramstogether with performance datagathered over a longerterm canbe the basisof decidingwhich areato tackle first. In this casethe line operatorsare largely concerned with monitoringraw materialand finished productto respondto regularmachine operation faults, if monitoringis unavoidablemachine operationitself shouldbe monitored.
(ii) Rigorous application of the IDEF3 method provided some quantitative information for analysisof systembehaviour. It was initially difficult to define the boundariesfor the analysis when taking the 'bottom up' 11DEF3approach. This problem was solved by constructing a 'top down' IDEFOmodel to give an investigativeframework for ]ODEF3modelling which has provided the basisof the methodologyshown in figure 6.5. The method proved to be a rapid meansof structuring the necessaryinformation. It is conventionalto describea processas a sequenceof activities and it was noticeableduring the interview stagethat BDEF3lends itself to capturing information in this way. Gathering information for the open format of the
110 Elaboration documentproved time consuming. The format is generic and it is possiblethat a more defined structure for Elaboration documents in specific caseswould avoid missing information and could speedup the process.
Methodologyused in ca study Complete methodology ODnstruct'ASIV IDEFO framework Implementabon Fj jiý ýAS IS behaviour alidate'rOBI e IDEF3wt usi g IIDD 3 S0 IDEFOfra mieworkI Recommeridabon Figure 6.5 The methodologyto use IDEFo and IDEF3 diagrams The casestudy highlighted areasof further work to examinethe possibility of relating IIDEFo ICOM structure to an 11DEF3elaboration documents or PFNs. 6.4 STAGE 2- FORMAL REPRESENTATION OF A COMPOSITE MODELLING METHOD - IDEFO-3 The IDEF3 method has been developedto capture a representationof a process centred view of systemoperation. It can representprocesses (UOBs) connectedby temporal, causal and logical relationshipsin a ProcessFlow Network (PFN). It is this capability to describe aspectsof systembehaviour, together with IDEFo functional modelling capability, that forms the compositebehavioural approach EDEFO-3. Mayer et al. (1992a) recognisedthat there is a relationship between the 'activities' in an IIDEFomodel and the 'Units of Behaviour' (UOBs) in an IDEF3 process flow description. They acknowledgethat the IIDEF3 method was designed'with interaction in mind' and that 'the BDEF3syntax recognisesthe relationshipby providing a meansof referencingassociated ill IDEFo activities from within an IDEF3 UOB'. They also define a UOB as a 'neutral term' representinginformation about an event, act or process. Whereasthe IDEFOmethod uses Functions to describe'collections of related activities' clearly both a UOB and a 'function' can representthe sameaspect of a system. All UOB boxes in an IDEF3 diagram have a field to provide a referencenumber to an activity in an IDEFo model. Benjamin et al (1993) have proposed an integrated approach to utilise both methods for activity based costing applications. Similarly in a US Defence Logistics Agency funded project Cesaroneand Dobrzeniecki (1992) advocate a hierarchy of both methods in the design of the 'next generationof manufacturingsystem' however no methodology or modelling syntax has been developed to correlate the two methods. Both methods produce models in a class that Ackoff (1962) defines 'symbolic' as models that only -take on meaning 'when the symbols and things they represent are defined'. Central therefore to any correlation of the two methods is a structured analysis of relationships (in terms of symbols) between IDEFo 'Functions' and IDEF3 'Units of Behaviour' (UOB s). The relationshipsbetween model elements and modellingconcepts or syntaxused in both IIDEFoand the ProcessFlow Network (PFN) of IIDEF3are presentedin graphicalform in figure 6.6a andb. This graphicalapproach is an alternativeto the algorithmicapproach to analyseIDEFo and IIDEF3 model structure used by Kusiaket al. (1994). Figures6.6a and b describe'therelationships between the graphicalbuilding blocks (symbols)used to describe the real world or tangible elementsof systemsand the conceptsused to describe relationshipsbetween those elements. Figure 6.6a showsthe fundamentalsof IIDEFomodel diagrams. The IDEFOmethod uses diagramsto describethe relationshipbetween two tangible(real) elements of a system: 'Entities' - information,objects, material or dataused by an enterprise. 'Functions'- carriedout in an enterprise. The BDEFOmethod uses descriptivedevices in diagramsto representthe relationship betweenthe two tangible elementsof a system. In order to do this modelling 'concepts' are used. 112 KEY-. Concepts Tangible modellin elements Inputs, Controls, nposition Figure 6.6a A meta-modelof EDEFo Aggregation, Decomposition, UOB Precedence, Sequencelogic Figure 6.6b A meta-modelof IIDEF3 Participants in the system would not necessarilyrecognise them as concepts, they are the descriptive devices used to represent relationshipsbetween the two basic elements. For example the three conceptsused to relate entities to functions (i. e. the three relationships that can be describedby IDEFo) are Input, Mechanismand Control. Similarly: " the conceptused to relate functionsto entities is an Output. " the conceptsused to relate functionsto functions are Aggregation and Decomposition. " the conceptsthat relate entitiesto entities are Branch and Join. 113 Similarly figure 6.6b shows the correspondingrepresentation of an IDEF3 Process Flow Network where the one tangible element describedby the method is a Unit of Behaviour (UOB) that can be related by the concepts: Aggregation, Decomposition, Precedenceand Sequencelogic. Controls, Aggregation, /Aggrcgaý? n, Input ptes Decomposition Output glk\es *IF )SequenceAggregation, - Decomposition, UOB Precedence, logic Figure 6.7 A meta-modelof IIDEFO-3 A develop correlation of the two methodsto a composite approach- IDEFO-3, is shown in figure 6.7. New concepts are introduced to relate the three tangible elements (function, entity and UOB) used by IIDEFoand IIDEF3 respectively. In IDEFO-3 an IDEFo function becomesa context or 'scenario' for decomposition(the linking concept) to IDEF3 UOBs in a process flow network. Converselyaggregation is the concept used to take a functional view of a UOB in a processflow network. To completethe IDEFO-3 structure and correlate IDEFo entities to IDEM UOBs the liking concept of input and output 'gates' is introduced. Gates are used to identify and correlate inputs, outputs, mechanismsand controls from 114 IDEFO scenarios to UOBs in IDEFO-3 process flow networks diagrams. To enable relationships between ICOMs, process precedence and process sequence logic to be identified wMst maintainingconsistency in decomposition. 6.5 STAGE 3- AN APPLICATION OF EDEFO-3CONCEPTS The modelling conceptsdeveloped in stage I and 2 have been applied in a casestudy. The study takes place in a one of the URs major suppliersof health care producte. Due to the high volume production, salesturnover and retail competition, stores control is an order winning criteria for the companyand thus knowledge of is operation is critical. Figures 6.8 and 6.9 are context and viewpoint diagrams,nodes A-0 and AO respectively, of a model being used to examinethe operationof a raw material and finished product stores. Node AO 'Run stores' comprises three activities: Receiving material, Controlling material and Dispatching product. The diagramsrepresent phase I of the IDEFO-3 methodology (shown in figure 6.12 and discussedin section 6.6). A sequenceof activities is not indicated by the positioning of activity boxes in the diagram. In practice all three activities are carried out concurrently, the activity boxes could be positioned in any order and providing the correspondingarrow structure is maintainedwould be equally valid. In an'IDEFo diagram no logical relationship necessarilylinks two activities, each activity box is functionally independent. In this case study examination of the staff or equipment used to carry out activities was not consideredessential to the investigation. It was therefore decided that 'mechanisms'i. e. the staff or systemscarrying out activities, would not be included in the model unless it was of particular importance to the analysis and so no mechanismswere subsequentlyused. The 'Receive material' activity was of particular interest to the company and has been used as the basis of the casestudy. The IDEFOnode Al, in figure 6.9, has provided the scenario for decompositionto an IDEF3 PFN (figure 6.10) and subsequentlyto the IDEFO-3 diagram shown in figure 6.11. SchollConsumer Products, Derby 115 I- 0 9D EEj2 e > (L) E 2 CL ' @ Z0 0 (0 e0 1qoý3 1 0 U- n_ rn00 C w oo %0 mm U) 0 0 ,0 0 00 tr E rý a %D co 0 Z, UM D to m CO2 cg 012 4J c ý0 E- LL 6 O D3 Figure 6.8 Node A-O 'Run stores' 116 C" a to 0 9) E > 1:; E 0 9) 0 - L» Z 2 0 0 . m -U0 -a *0 (1) M I (n le: uU (D E 0 0 0 00 0A ft U\ cr i ID EO= or- A m ýE- 5 PL 20. 0 1-. ý: ., 0 ' 2 0 0 .,. .. 0 a. m CL E CD 9b to "-> IM 0 L' > m 0 m xýObub e4 0 Z rA (L) ý0< 20 1:ý I C 20:3 Figure 6.9 Node AO 'Run stores' 117 In figure 6.9 controls on activities are described,using IDEFo, by arrows entering the top of activity boxes using nouns to indicate the objects or information that govern the activity. For example in node A3 the activity 'Dispatch product' has two controls, 'Goods release book' and 'Delivery documents'. The activity is 'constrained' by the two controls together with the input 'Manufactured Items'. The circumstancesunder which controls act are not explicitly described in IDEFo i. e. are they required sequentially?, are they mutually exclusive?,are they complementary?,or how do they influence the process?. The IDEFO-3 approachcan help to clarify this. The next step is to build an AS IS IDEF3 PFN from the IDEFo framework. This is done by grouping the UOBs (or processes)into the activities identified in the IDEFo diagram. Figure 6.10 is the PIN describingthe sequenceand the logical relationshipsof all the UOBs in the 'Receive materials' activity. For examplethe first processis 'Accept material' followed by 'Identify priorities' followed by two processes either, hold items in goods receiving or processitems, the existenceof these alternativesindicated by the fan out asynchronous'OR' junction J2. FinaUyto correlate the IIDEFoscenario with the IDEF3 PFN an IIDEFO-3diagram is derived by consideringthe fbHowing: Fan out OR and Exclusive OR junctions represent potential alternative processing paths. They therefore implicitly involve control or decisions dependent on the availability of material, data, information or objects to facilitate, to allow or to choose a processsequence path. In figure 6.9 the IIDEFoNode Al, 'Receive material'. has three controls 'Quality specifications', 'Priority book' and 'Purchase orders'. Using the BDEFO-3diagram (figure 6.11) the three controls can be correlatedwith the four fan-out junctions J2, J4, J9 and JIO. JunctionJ2 involveddeciding whether to 'Hold items' or 'Processitems'. Goodsinward staffused a 'Priority book' asa controlthat simplyidentified those part numbersand in some casesquantities that required immediateprocessing. lis Figure 6.10 IDEM PFN - 'Receive material' 119 JL 7 006 (5 p 0 CL CL IZ :3 0 2 IL CD .,4 V cr ce 10 k. k) 1-MD1 MA A-b 4- "ri 0 9 I 44 -;ý *G; "0 0 u > ii Figure 6.11 IDEFO-3 diagram 'Recebýematerial' 120 Junction A involved a decisionto; either except the counted quantity and produce a goods received note (notifying Quality Control (QC) staff that parts or material were awaiting inspection)or raise a discrepancynote to sendto the supplier. In order to make the decision a copy of the purchaseorder was necessaryto define, as a minimum, a part number and order quantity. JunctionJ9 involveda decisionby QC staffto acceptor rejectparts or material.If theywere accepteda concessionnote could be raised(as a result of a decisionindicated by junction JIO) if someaspect of the item did not conformto specificationbut was stiff acceptable.For thesedecisions the only documentedcontrol usedwas the part or materialspecification that couldbe a drawingor text description. Figure 6.11 also Mustrateshow the open endedprocess paths of the PFN (in figure 6.10) can be correlated with the inputs and outputs of node Al (figure 6.9). The outputs of IDEFo node Al are; * Goods receivednote Material Concessionnote Rejecteditems & documents Discrepancyletter to supplier All of which areproduced as the result of the from the five processpaths that terminatethe PFN in figure 6.10. The IIDEFOdiagram therefore correlatesin two domains,IDEFO outputs with the 'open ends' of the PFN and IIDEFocontrols with separatestages in the IIDEF3PFN. Similarly the single IIDEFoinput to 'Receive material' (node Al in figure 6.9) is 'Purchased items'. 'Purchaseditems' are objects necessaryfor the first process(Receive items) of the sequenceof processesin the PFN of figure 6.10. The complete EDEFO-3diagram is presentedin figure 6.11 showing the correlation of input, output and control structure from the IIDEFOdiagram with the IDEF3 Process Flow Network. 'Gates' (discussed in section 6.4) are used as the symbol representing the boundary between IDEFOand IDEFO-3 diagrams. For the purposes of this case study an 121 IDEF3 PFN (figure 6.10) and IDEFO-3 diagram (figure 6.11) of 'Receive material' were produced. In practice flis is not necessary,IDEFo diagramscan be directly decomposedto IDEFO-3 diagrams. 6.5.1 A review of stage 3 The approachand resulting model significantly contributed to the analysis. Suppliershad, in the past, been receiving both goods return and discrepancy notes separately from the companyrelated to the samebatch of componentsor material. The effect over the long term was; a lack of trust between company and supplier, the need to deal with two company departments(Goods inward and QQ to resolve problemsand the cost of material shortages and administrationwhen communicationswere not timely. The IDEFO-3 diagram (figure 6.11) demonstratedthat the three decisionsinfluencing UOBs at J2, A and J9 were not co- ordinated and the following points revealedby the model were the focus of discussionwith the managementstaff responsiblefor re-structuring: 1. Once Goods Inward staff had acceptedmaterial, split batchescould occur as a result of decisions at J2 which could result in discrepancy notes being raised. This. was particularly evident after shift changes. Counting and Inspection were sequential processesinvolving two separategroups of staff actingon separatedecision criteria and could in somecases involve removingand replacingmaterial on palletstwice before acceptance. 3. After acceptanceby QC staff (UOB-7 in figure 6.11) items were assignedto store locations and in some cases concessionnotes for production were raised. No documentedcontrols, proceduresor guidelineswere used to decide whether a concessionnote was raisedand this lack of control meantin most casesconcession informationwas not sentto suppliers. 122 Points I and 3 were made apparent by the IDEFO-3 approach and could not have been explicitly describedby building IDEFO or IDEF3 models in isolation. Point 2 could have beendeduced from building an IDEF3 model. In practicedocumenting and modelbuilding is an iterative processand it was found that althoughthe IDEFodiagram was a necessarystarting point it was refinedas the IIDEFO-3 diagramwas developed.The bottom up approachused to group sequencesof UOBsfor the IIDEFO-3diagram forced the analystto examinethe criteria used to selectsequences of UOBs which in turn highlighteddeficiencies in the correspondingIDEFo diagram. Sirnilarly an open endedprocess path in the PFN forces the analystto considerthe validity of the outputsin a correspondingIIDEFo diagram. 6.6 CONCLUSIONS For modelling methods to be useful in manufacturing enterprises they must provide an analyst with a meansof understandingand solving problems by either gaining insight into existing systemsor contributing to the design of new systems. The case study using the IDEFO-3 approach proved how critical it is to provide an insight into the behaviour of systems. The IDEFO-3 model required fewer levels of decomposition in comparison to typical IIDEFomodels of similar situations (in the authors experience). IIDEFoand IIDEF3 diagramsvalidate eachother and ICOM gatesin an IDEFO-3 decompositionforce an analyst to examinethe consequencesof functional 'Control' in terms of processlogic or decisions and the significanceof processsequence in terms of 'Outputs'. The approachuses robust, well understoodprinciples and a commonly acceptedset of diagrammingsymbols whose use is alreadyunderstood by analystsand engineers. For the IODEFO-3approach the objectivesof an analysisare defined in the purpose statement, as in a conventional IIDEFo model. The statement along with the IDEFo context and viewpoint diagramsare the initial steps in the modelling process. Gathering the necessary information for model building is conducted using the approach described in the IDEFo manual (US Air Force, 1981a). The results of which are representedas a seriesof IDEFo 123 diagramsdecomposed from the viewpoint diagram. A functional analysisis thus the first phaseof the methodology. The casestudies suggested that it is easierand quicker to establisha commonunderstanding of the essentialactivities in an enterprise using IIDEFoinitially than to take a processview and deal with the detailed logic of process sequences. This is in contrast.to Mayer et al. (1992a, p.4) who suggestthat IDEFo models can be constructedafter gathering information in IIDEF3models. The IDEFO-3 methodology requires consideration of the transition from IDEFo format to IDEFO-3 diagrams. This should take place at the stage that functions in IDEFo diagrams become scenariosfor IDEF3. process flow networks. Developing the graphical aspectsof the IDEFO-3 approachallows balancing the need to model as many important variables as possiblebut still maintain useful, but simple, descriptionsthat provide an understandingof the relationshipsbetween them. During information gathering it was found that, as domain expertsexplain situations,the finer the level of detail the more likely their description moves away from describingsimply the functions that occur. At this stage an expert's narrative is often the logical flow of activities that supports representationas a process flow network. This finding echoesMayer et al's (1992a) assertionthat 'experts expresstheir problems in terms of an ordered sequenceof events'. The point at which it is useful to decomposean IDEFo diagramto an IDEFO-3 diagram is dependenton the actual situation being modelled and the use to which the model is to be put however three potential pointers are in evidence: (i) When decomposition of an (IDEFo activity) exposesan obvious or simple sequenceof dependentprocesses. (H) At the stage in analysisthat requires an understandingof how controls identified using IDEFo affect activities. - Are they all required, are they required simultaneously,what decisionsdo they influence and what is their influence on the logical relationships of sub- activities. 124 knowledge when the of domain experts was expressedas a waterfall -a sequential descriptionof processes,activities or eventsusing their relative positions in the sequenceas a frameworkfor discussion. A key to using the approach is correlation of IDEFo and PFN levels of a model using an EDEFO-3diagram to examineall fan out junctions and to identify the presenceof a control in the parent function (scenario). What constrainsthe decision or actions necessaryto follow any of the subsequentalternative sequencesof UOBs. If no constraint is evident that point in the UOB sequencerequires further investigation. If open endedflow paths that start and endthe processflow networkdo not correlatewith IDEFOinputs and outputsthis pointsto an investigationto seeif thereare unnecessary, redundant or missingprocesses that haveno influenceon the inputsor outputsidentified at the IIDEFolevel. Carryout IDEFO 'AS IT fl=tional Analysis Model 70 using IDEFO-3 UscIDEFO laccnarios, asbasis of IDEFO-3 PFN MEMODOLOGY C4rlstructi(m Rocommend 70 BE Coutrixt IDEFO-3 &grams Awlysis Figure 6.12 The IDEFO-3 methodology 125 In parallel to diagram correlation, validation of the accuracyof analystsview of the systemis carried out using the author reader cycle in the usual way (US Air Force, 1981 a) (although in the casestudy, validation was limited to the expertsinvolved). After correlation of IDEFo and the PFN the resulting IDEFO-3 model can be used for analysis, followed by recommendationsfor changeand the model has the potential to be used to describethe TO- BE proposalsbefore implementation. The complete methodologyhas evolved into the form illustrated by figure 6.12 as a single cycle. An outcome from the case study was the developmentof an 11DEFO-3methodology user guide (reproducedin appendix2). It provides a guide to the method and syntax associated with phase2 and 3 of the IIDEFO-3methodology i. e. Use 11DEFOscenarios to construct PFNs and construct IDEFO-3 diagrams. In the case study the Design ADEF (MetaSoftware) tool was used to produce IDEFo diagramsand IDEF3 diagramswere produced with conventionalgraphics software. At this stagein the developmentof the approachthe lack of software facilities for IDEFO-3 such as syntax checkingand symbol configuration is a disadvantage. The developed IDEFO-3 approach can describe or represent functional and -behavioural aspects of a system for cells 50,60, and 70 of the manufacturing enterprise modelling framework (shown in figure 6.13). It is not designedto support a dynamic perspective representedby the cells 40,80,120 of the manufacturingenterprise modelling framework. A more extensiveapplication follows in chapter 7 to re-enforce the findings, develop some of the featuresand demonstrateadvanced aspects of the approache. g.: At what level in decompositionthe transitionfrom functionaldiagrams to a PFN is most effective. How effective the approach is in a higher level manufacturingsituation when considerableaggregation is necessary. * The detaileddiagram syntaxassociated with IIDEFO-3. a The suitability of the approachfor TO BE models. 126 * Its usefulnessin a larger scaleproject. 9 To demonstratethe approachcan be usedby staff other than the analyst. * To validate the IDEFO-3User Guide. FOCUS a b C CONFIGURATION BEJ1AP7OURAL QUANTITATIYE What,WheM Who How,When HowMuch. How PERSPECTTW toaickly, HowMan I ENTEMPRISE 10 50 90 Values, IDEFo XDEFO-3 GoaKScope ID-E-FO-3 2STRUCTURAL 20 60 100 Functional IDEFo XDEFO-3 Relationships, IDEFO-3 Interfaces 3 PROCESS 30 IDEFO 70 110 Thestages involved in TDEF3 1DEF3 informationand F0-3 XDEFO-3 materialflow -1DZ. 4DYNAAHC 40 so 120 Theactivation of informationor object now I ENTERPRISE Product,Material, Constraints,Material Resourceconsumption, OPEAWON Resources,Locations, flow, Schedulesand Output.Perfwaance Information. timing. Figure 6.13 Mocation of IDEFO-3to the manufacturingenterprise modeffing framework 127 CHAPTER 7 VALIDATION OF THE IDEFO-3 METHODOLOGY 7.1 INTRODUCTION This chapter describes the validation of IDEFO-3 through its extensive use to design a manufacturing system. A case study is reported where the methodology contributed significantly to the successof a project to design a manufacturing system for fabricated component manufacture. The major UK manufacturingcompany where the case study was conductedis in the competitive global environmentdiscussed in section 1.2. To survive and developthe companyis using a combinationof technology, organisation,management and the control of information and is demonstratingits ability to adapt to a new situation. The enthusiasmwith which modellingwas acceptedby the staff involved is a reflection of the open approach encouragedby senior managementin an effort to explore every potential route to gain a commercialadvantage. The company was selected because it contained many of the features that currently characterisechange in UK manufacturingsuch as; the implementationof processautomation, the need to use computers effectively, the need to reduce direct labour costs and changea legacy of operating practicesand principlesbased on local knowledge and manual operations. Modelling was used to analyseand design the structure, operating principles and software requirementsfor a new manufacturingfacility. The companyalso provided an opportunity to validate the IDEFO-3 approach, testing its functionality, capability and ease of use as a methodology to designa large manufacturingsystem. 128 The preliminary work covered modelling both systemconcepts and detail systemdesign and took place over the period from November 1994 to May 1995. Other complementary approachesused in the analysisphase (not reported in this thesis) included production flow analysisand discreteevent simulation. 7.2 THE CASE STUDY SITE The case study was carried out in collaboration with the Dunlop and Rankin Steel Service Centre of British Steel Distribution (BSD) in Leeds. The centre is one of sixteen UK distribution centres for 'General Products' typically grade 43A plate, floor plate, beams columns and joists, hollow sectionsand bright bar-stock. The 200 acre Leeds site is used both for steel-stock storage and has the capacity to shear, profile, cold saw, shot blast fabricate and'paint steel-stockto customerrequirements. The site has beenused to provide a distribution service for the past twenty years, however the businessobjectives of BSD have changed significantly over the last ten years and the need to process and add value to the steel-stockis now consideredthe key to commercialexpansion. Toronto Engineers Ltd. was establishedeighteen years ago as a fabricator for the mining industry. The collapse of the mining industry and a failure to diversify led the company into receivership before it was taken over by BSD in 1990. Re-named Toronto Industrial Fabrications(TIF) Ltd. and re-locatedto the BSD Leeds site, it is now a 5000m2 generalsub- contract fabrication unit for profiling plate and welding sub-assemblies.The successof the company has led to a long term commitment from a major manufacturer of earth-moving equipment for the supply of 'kits' (which are palletised sets of fabricated and profiled components). In order to achievethe quantity, consistent delivery schedulesand quality of product demandedby customersBSD committed a 12 million investment for new plant and systemsto expandthe TIF workshop, equipmentand staff The decisionto proceedwith the investmentwas taken in March 1995 and the first orders for mechanicalhandling equipment were placed in April 1995. 7.3 THE PROBLEM DEFINED The basis of the casestudy is a proposalby TIF to supply a fixed, daily quantity of fabricated kits to an earth-moving plant manufacturer. The BSD manufacturing facility (named the Construction Earth Moving Equipment (CEME) Centre during the developmentstage of the 129 project) was to be an element of the customer's JIT supply chain. Businessperformance, product quality and scheduleadherence were critical to the successof the unit and were the overlying considerationduring design. At the stageof order acceptanceand a commitmentto supply kits by BSD, technical customer information comprised a proposed twelve month master production schedulefor six kit variants and a complete set of assemblyand detail drawings (modelling applications concerning refinements to the commercial relationship betweenBSD and their customersis not consideredin the casestudy). The e)dsting TIF fabrication unit was a traditional jobbing shop supplying a range of customerswith batch quantitiesof profiled plate and welded fabricationsranging from five kg to eight tonnes. The key processeswere gas and plasma profiling with subsequentmanual fettling, shot blast, manually controlled bending and piercing and manual welding. A small amount of machiningwas sub-contracted. Amongst the range of componentsmanufactured, a 'Contract' unit processed7000 tonnes of plate annually to produce the kits for earth- moving plant manufacturers. Problemshad been experiencedwith late deliveriesand there was a history of kits being returnedfrom customersdue to qualityproblems. The manually operated shot blast, piercing, rolling andbending equipment was approaching the endof its effectivelife andcontributing to the quality problems. All kit componentswere profiled from standardplates and the unit reliedon manualprofile nestingand gas profiling for this first processstep. Bottleneckshad becomeincreasingly common at the profilingstage as production levels rose. This situationin the fabricationshop together with the shift in businessobjectives to expandplate processing to 20,000tonnes annually was the justification to embarkon the CENIEproject. Planning for the CENIE centre was initially focused on plant layout and mechanicalhandling equipment, which is where the expertise of the existing staff lay. Capacity planning for manufacturing was largely overlooked until the volume of plate being processed and the subsequentimportance of profile nesting together with the need for computer support for manufacturing were considered. When the significance of manufacturing capacity to the future of the centrewas recognisednew staff were assignedto the project and at that stage,in November 1994, the author becameinvolved. 7.4 A PRELIMINARY ANALYSIS The site had two closely related manufacturingunits the 'Spot' unit and the 'Contract' unit. They were located in separateworkshops with their own dedicatedmachines, tools, and shop 130 floor supervisionbut they had common managementand drawing office staff. The Spot unit was basedon a sub-contractplate profiling service. It serveda wide range of customerswith profiled plate in quantitiesthat could be in severalhundreds but were usually lessthan twenty- off per order. Demandwas consideredrelatively unpredictableand componentswere profiled with few subsequentprocesses being carried out on the profiled components. A decisionhad beentaken to considerseparation of the Contract unit (to becomethe CEME centre) from the Spot unit before this researchwork began. Initial contact for the case study was made with BSD's engineeringoperations manager and after some discussionsthe companywere convincedof the need to model and plan the change particularly in view of their relative inexperiencein the manufacturingenvironment they were considering. Manufacturing control in terms of scheduling,material flow and processeswas of particular concern. The case study was carried out using the methodology describedin the IDEFO-3 user guide presentedin appendix2. The case study was initiated by considering the Manufacturing Enterprise Modelling Framework in the context of the initial phase(Carry out IDEFo AS IS functional analysis)of the IDEFO-3 methodology shownin figure 6.12. In order to understandthe existing system,to analysethe situation and to provide a basis for re-configuring the systeman AS IS IDEFO-3 model was generated. The context boundedby cell 20 of the manufacturingenterprise modelling framework shown in figure 6.13 provided a context for use of the model, i. e. a structural perspectivefocused on the configuration of the systemusing IIDEFodiagrams. It was decided, during the initial discussions,that aspectsof control such as staff reporting and structures, budgets and commercial relationships with suppliers and sub-contractors would not be part of the core modelling effort. The modelsused in the casestudy reflect this approach. A pre-determined (at the stage that the author was involved in the project) decisionto proceedwith the CENE centremeant that enterpriseperspective models (cells 10, 50, and 90 of the manufacturingenterprise modelling framework) were not of much value at the stagethat modelling commenced.The goals and scopeof the project had been established and the problem facing the companywas how to proceed with the change. The enterprise perspectiveis not referred to againas the project continues. 131 7.5 THE AS IS IDEFO-3 MODEL OF 'RUN SPOT AND CONTRACT MANUFACTURING' 7.5.1 Phase 1 of the IIDEFO-3methodology IDEFD-3 Cany out IDEFo AS IS functional analysis METHODOLOGY rn.,.. 1 '7 Interviews with the General manager, Manufacturing manager, Commercial manager, and project engineerswere conductedto derive the AS IS model shown in figures 7.1 to 7.10. The model starts with figure 7.1, the Context diagram (node A-0, Run Spot and Contract manufacturing)defining the 'Purpose' of the IIDEFO-3model 'To investigatethe operation of spot and contract departmentsat BSD Leeds'. The 'Viewpoint' of the model reflects the immediate priority of BSD, that of manufacturing control in terms of material control, schedulingand the relationshipbetween manufacturing processes. The diagrams node AO, Run Spot and Contract manufacturing, figure 7.2 exposed the functional structure of the units. An analysisof node AO, revealed the autonomy of shop floor supervision in the two units. 'Spot schedules' and 'kit schedules' (highlighted) comprisedof a parts lists in alphabeticalorder and a verbal statementof weekly quantities and customer due dates for Spot profiled parts and kits which left shop floor supervisors responsiblefor assigningpriorities. The diagram also revealeda heavy reliance on heuristics and verbal instructions, a situation commentedupon by local managementwhen the draft versionsof the diagramswere discussed and validated. The inter-dependenceof the two unitsin termsof profilingcapacity can be analysedfrom the diagrams.Under-capacity in Contractunit led to 'Profilingrequests' to the Spotunit. Similarlyunder-capacity of fettlingor shotblast processes in Spotmanufacturing resulted in 'Profiledcomponents' being fed to the Contractunit for finishing(shown in figure7.2). In additionsupervisors in the two unitswere not explicitlyaware of eachothers schedules again this is madeclear by the diagram. 132 1 4) . E 0 190 .& so C) ig . .0E0 !R mr c to * Co 0c0 00 E Je m (D E (0 Co 3.2 w" i 0 C bý M (09 cn e0 W .2 0 m 0bU r_ 12 Cw E r v 00 0 CL *0 t mom r_ 0 0 Co m AL ;p0 19 Z 20 Oug 0 CL 2.2 (D , CL c) 0 06 m 92. cý 00 o(0 Z t- ý0, 912 c Co 2 rA2 en 9 12 0 m i2 0. c Co f0 Z cE 0 9 K 03 JR CI) 0 g! gIZ , p4 0 2 0E E Z CLJM 0 m m 0 ij ci 0 E:i IL Figure 7.1 Node A-0 Run spot and contract manufacturing 133 jo (b 1 o JR CC, r c CL m 0c0. . E : (1)E E ? 8-0C.: 0 oý 12 C.) Co0 (D 0 E m c4 0 gAL F.: 0.E e co AL to0 M 40 - c 22 Z% 0 09 ät to ecy02 CL. E .1 4-, 0 9 r ' 1 Co 4 CYCi CL 0 0 0-£2 --7 toi', gl. 06 -e 00 a e o E i, ' l 1 1 0= Z -\ 4- e ý m - Co .., - - 124a g Co- " is 3:l! mZc(0 0 to 0 (bk-E (0 0 00 00 -42 0) rý m@ a rn 8 Z- i %D l Z, 0 w r4 ID E '0 Z ID 0 2 CI) ID M w 0 ll c 0CL E x0 Z (D U) m E , -am -a WC; 2 r- E l! Co tu Z 0r _. iR 0r_ 0Eam 0 r_ w A0 co Figure 7.2 Node AO Run spot and contract manufacturing 134 0 A 0 jý r» 0 E E0 (D «a w 0 0 Co m -0 e r C 0 Co 2 r6c . J l' E ID 0 ß r0 < r_ CL 0 0 r_ -0 > 'DI 8 (D t- CY E E ID - r- 0 0 Je A) cm Z 0 m P0 r_ n r% CD *0 ty Co vi 9 %D -2 0=1 CD E (1) 0 %- m (D A E cm Er ein L :e. E l! 1 - Z28 Z 0, (D 10 e 'CJ . ci Z rn rn 0 r_ i2 0 IR le0 ý0 12 f- CL la -. ro e :2 cr 0 1 Z0) CL. .Z Ir CL v2 :D 0 L9 Figure 7.3 Node Al Manage manufacturing 135 0 r CD E E Er 0 Z 0 cy 0 3 (0 w E Z A e CL (0 .0m 0 QwZ 0Z CL w 75 8 .e9 CL 0 ID CL 0 0. CO ID lu 0- 9. .00 r. (D 0 Z (0 02 As 0% - CI)0 < c: 00 U E - c tue CLa x 1 0 2.0 : E 0 rL , 1 e4 tu E 0 (D 0 E 0 Z gl) O =. c ' t Z Figure 7.4 Node Al I Establishplate, capacityand componentrequirements 136 U) E 0 (D C) 0.. 0 V 00.q2 ta W 0 CL 'a M6 CL :33 I 3 [a 0 to (h c J,D 0 in JR.2 g gC 00 1! vM 8 en cc I M co E g to 3.0 0 0 Figure 7.5 Node Al 13 Processcontract orders 137 v 0 Im,16 m W ra A r. A CL 2' N 3 1 12 .90'5. (Dtr 2 tg U) a) o 094 to Ix 4) IL IL U) 3 :,a: tn Q c , "D 21 E $A Q AA2 C rn LJ ca z a P0 ý o .9 1 10 2 iq, R i 0 4D Ar. r_ O'mw 22m I 3 x 'a 4 ý, _ (D E 3 ig cr E -o r. AL LD CD CL 9.9 . 0 b- . 0 _0 V) o %D a .0 CL i- M (/) IE A e4 0 0) ,ý i (D M 0 :9 E RL w Li 0 C0 0 0 co J3 Q -0 M fu cE a w e E o o 0.0 a. 0 Figure 7.6 Node A3 Run contract manufacture 138 ui (1) 0 W (D CL ui so U)4) cL E 7E. ,Z-- -0 E E 0 10 0C) r0 1 Je8 Co LO -- C) lt 0) 1 C I (ft Co C .0 2 ýi-, 0 r 0a.! tdim. .5 1- IMO 0E (L) At .Zmr r- :3 9 21 0X tr 0 0 72 U) c 0 CL 0 oE LU - zic a. ý Ca Cla Z------ý ...... 0 0 (0.5 I W. 'a_j Figure 7.7 Node A03-32 Producework to lists 139 A AA Xl Je F18 20 0 10 1'D rEL 0 2 0) a I 'rD ß_ Frfoui .rM cl 0---5 U) 0 9 5- IC29L u) 92. U) o x 3. 0 i -0E U- p§ C- CL 17 1A Lo to ------1ý- I M. X CO I i (D vc ý2 CL CL :p 122 A EEO 2- CL ? L C) E .21r,2 Cr CL 8 x Figure 7.8 Node AOi-33 ManufactureKits 140 o U) 0) Ia. 4) E CLch 15 M - e ,- g.0 r: Z ri m C=01-q rb S 41) c m E C"s cI Figure 7.9 Node A03-325 Prioritise components 141 In validating nodesAO and A-0 and discussingthe situation with the staff involved it becameevident that this long acceptedpractice was the causeof nervousnessin contract manufacturing schedules. The model revealed that shop floor reporting to operational managementwas based on exceptions and that performancemeasures at shop floor level were simplistically based on finished items (dispatch notes were signed and fed back to administration as kits and components were loaded on transport for delivery). Exceptions could be capacity, breakdown, material shortages,bottlenecks, etc.. The model showedthat scrap, re-work and detailed work centre performance was not monitored (apart from time and attendance information for direct wages). Analysis and validation of the model supportedthe proposal that the CEME centre be run as an independentunit comprising aspects of node Al and A3 with its own management, schedulesand manufacturing organisation. The successof the centre would depend on delivery reliability and modelling showed that shop floor schedulemanagement was a cause for concern. As a consequencenodes Al and A3 were decomposedfor more detail using interviews with shop floor supervisors. The decompositionsof Al and A3 are shown in figures 7.3 and figure 7.6. Figure 7.3 shows node Al 'Manage manufacturing'three functions were revealedand associatedwith Al: Al I - Establishplate, capacity and componentrequirements, A12 - Manage and produce drawings and A13 - Administrate unit. Node Al I 'Establish plate component,and capacity requirements' is the activity that balanced customer orders with shop floor capacity (staff and processes) and the inventory of completedkits and componentsto produce shop floor schedules. Using customer orders a plan of throughput (in terms of the weight of steel) processedan outline of capacityrequirements in terms of staff and processeswas established. Node A12 'Manageand producedrawings' was found to be largelyconcerned with producingdetail and fabrication drawings for Spot manufacturing.In additionit was found that the management,storage and reproductionof customerdrawings for Contractmanufacturing was alsocarried out togetherwith someinformal liaison with customersover designand manufacturing problems. Node A13 'Administrate unit' is the function that controlled the releaseof documents such as schedulesfor raw material (plate) requirements to BSD Leeds and the 142 ordering of purchased components, consurnablesand sub-contract manufacturing. The activity included the issueof dispatchnotes, entering data into the site accounting - system and dealing with other BSD administrative units such as Personnel and Finance. In fight of the 'Purpose' of the model 'To investigatethe control, organisationand operation of spot and contract manufacturing at BSD Leeds' and the manufacturing 'Viewpoint' sufficient information had been gatheredabout nodesA12 and Al 3, however, node Al I was decomposedfurther to expose more detail about the control and operation of shop floor scheduling. By an analysis of node Al 1, Establish plate capacity and component requirementsthree activities were revealed: Al II- Review orders, Al 12 - Process spot orders, Al 13 - Processcontract orders (shown in figure 7.4). Node Al 11, 'Review orders' was an activity carried out by senior managementand marketing staff. In the activity Spot and Contract orders were treated as separate items and no attempt was made to reconcile capacity in the two units. The activity was largely concernedwith prioritising orders to achievean aggregatebalance of load on resources. The outline scheduleswere expressedat this stageas the number of kits or the weight of profiles per week for specific customers. Priorities could be influenced by a complex mix of customer priority, material shortages, shop floor manufacturingproblems and managementpractice. With the additional constraint of a material flow policy of profiling batchesof components,with common plate thickness, in groups. Node Al 12, 'Process Spot orders' was the activity that produced detailed schedules and plate requirementsfor profiling Spot orders. Node Al 13, 'ProcessContract' orderswas the activity that balancedthe aggregate schedulesfor kits from node Al II with a knowledgeof the availablecapacity and capabilityof the plate profiling equipment. A schedulefor downstreamprocesses after profiling was not considered. The activity was carried out by production plannerswith a detailedknowledge of the gas and plasmaarc profiling machines. Componentnesting and plate utilisation was consideredto be of primaryimportance. The experienceof the productionplanning staff, much of it not documented,and standard'nests' (to providemaximum plate utilisation for existingcustomer profiles) had a largeinfluence on schedulegeneration. As batchesof profileswere produced they were transportedto the next process,the profiling nestsand schedulesthus becamethe driver for all subsequentprocesses. This activity alsoproduced a demand 143 pattern for raw material (i. e. Steel plate in various gradesand thickness)and produced requisitionsfor sub-contractprocesses. The importance of this activity to the investigation prompted decompositionof node A113 to the next level of detail (figure 7.5 Node A113, 'Process contract orders') revealing three activities: Al 131 - Review Idt schedule, Al 132 - Negotiate plate influence supply, Al 133 - Negotiate with sub-contractors. This diagram revealedthe of independentconstraints on the generationof Idt schedules.Developing the diagram exposed that, in practice, 'plate supply options' was a major constraint on the generationof schedules. Figure 7.6 is a diagram showing node A3 'Run Contract manufacture', four activities were lists, exposed in the analysis: A31 - Monitor inventory, A32 - Produce work to A33 - Manufactureldts and A34 - Monitor shop floor activity. basis input A31 - Monitor inventory, this activity was carried out on the of and output control. Profile quantities received from Spot manufacturingwere logged in to the Contract unit. Plates from BSD (or other suppliers) were received through acceptanceof a goods received note. Consumablessuch as gas, welding spools (for Mig welders), abrasivedisks for fettling, etc. were also received through acceptance of a goods receivednote. The output from the systemin terms of kits was monitored using dispatch notes that were completed as kits were loaded on BSD trailers for delivery to customers. Requisitions were raised to replenish consumables,and for plates suppliedby BSD. kit A32 - Produce work-to fists, in this activity scheduleswere usedto produce fists of individual componentnumbers using establishedknowledge of previous batches and customer drawings, formal BOMs were not used. Batch sizes for each component were basedon lot for lot againstkit requirements. A33 - Manufacturekits, is the activity that describesthe sequenceof processes necessaryto make each componentand assembly. Work to fists were issuedto 'profiling' (the first processfor all components).As componentsleft profiling they were distributed to subsequentprocesses. Priority for downstreamprocesses therefore,was stronglyinfluenced by the priority at profiling (unlessprogress chasing intervened).No formal sequenceof processeswas usedfor eachcomponent, batch moveswere dependent on standardpractice and expediting. 144 A34 - Monitor shop floor activity, involved collating staff time and attendance records, to produce aggregate direct labour hours, work centre and staff capacity problemswere logged as exeptions. Delivery documentswere assignedto completed kits and individual components were expedited to collate kits and achieve kit schedules. 7.5.2 Phase2 and 3 of the EDEFO-3methodology: Use VEFD4 IDEFo scenariosto constructPFNs and klmoDou)oy Construct IDEFO-3 diagrams 7. I. " At this stagein modelling it was evident that the complexity of two important functions, node A32 and node A33 (figure 7.6) neededto be understood. A description of these functions was necessaryto develop proposals for shop floor control in the CEME centre. Further decompositionusing IIDEFofunctional decompositionwas not describingthe behaviour of the two critical functions and it was at this stagein modelling that IDEFOfunctions were used as the scenariofor IDEFO-3 diagrams. Node A03 - 32 and node A03 - 33 are shown in figures 7.7 and 7.8. These two diagrams exposed some important behavioural characteristicsof the contract manufacturing unit and the details are discussedin section 7.5.3. It was also found necessaryto decomposenode A03 - 325 the processis indicated on figure 7.7 as a shadowedbox and shown in complete form in figure 7.9. Figure 7.8 is an IIDEFO-3 diagram that showed that all components undergo three fundamental processes;profiling, shot-blast and 'process profiles'. In investigating node A03-338 (processprofiles, shown highlighted in figure 7.8) it was evident that the complexity of sequenceand logic of the sub-processesinvolved for each component would be better describedusing a part number and processanalysis. A part number/processmatrix was used to describethe possible alternative routes that could take place in 'process profiles' and is shown as an FEO diagram appendedto the model (figure 7.11). Figure 7.10 is an extract from the matrix, it shows ten of the thirteen possible processesthat a component might 145 undergo in the horizontal axis. Part numbersare shown in the vertical axis and processing times for eachcomponent are allocated to each process,. The matrix was subsequentlyused to classify componentsusing processflow analysis,a part number/processmatrix is shown in appendix 10. CEME Centre floor to floor processi P ess Activity Cycle A Times (S ec) No. off per profils, Profile ore" I Manual Robobc Part No. Class Turn. Shot Table Shol Send Chamfer Rq-work Total ci Ass Gas Plasma Chamf Id Weld IS-Oorw_ - - A 420 496 480 1398 Bess 9R-886&1 2 A 192 120 so 372 9R4ftM 1 B 30 30 so 9R-8866f3 I B 30 30 80 9R4kW4 I C 0 9R-886&5 2 C so 120 230 Top hat 420 1 529 3DO 1249 911-11514 88 120 60 268 9R-8514 94 1 1 0 2 0 904 IDR4514 26 60 3-0 118 9R-8514 18 1 60 1 30 1 1 +-! 06 2 11 1 1 --- 21 Figure 7.10 An extract from an analysisof alternativeprocess routes AO I A-0 Al A2\ A3 Al IZI 12 A13 A31 Al A2 ýN3 I Al 12 13 A33\A33 A34 Al 131 Al 132 A03 -32 Al 133 A03 -33 A03-325 Spreadsheet Figure 7.11 Node tree of 113EFO-3model 'Run spot and Contract manufacturing' 146 A node tree diagramfor the completeAS IS model is shown in figure 7.11 showing the depth and breadth of AS IS model decomposition. Cý.. ID&F, le 6-d W- 7.5.3 Phase 4 and 5 of the EDEFO-3methodology IDEFV-3 Key findings and recommendationsfrom the AS IS MEr-H--"ýODOLOGY IDEFO-3 model analysis C.. - IDlY1 I Analysis of the AS IS model exposed a range of potential problem areas and formal presentationsof the model to BSD staff were focusedon the following points: The clear needto separatethe two manufacturingunits. To provide an interface using the 'Manage manufacturing' activity (figure 7.2) so that any requestsfor capacitybetween the two units would be considered by activity Al I (figure 7.3) and could therefore be evaluatedon the basis of overall capacity and integrated into existing Contract or Spot schedules. No clear definitionof product structureswas available,bills of materialwere not used consequentlythe conversionfrom independentdemand for kits to dependentdemand for individualparts took placeat shopfloor supervisionlevel (activity A32, figure 7.6 and node A03-32 junction JI, figure 7.7). No start time off-settingwas carried out for componentsat different levels in product structures. As a result detailed capacity planningwas not carriedout and aggregatingbatches of similar componentswas only done in obvious casesby individual operators. Similarly the decisionsinvolved in junction B and the UOB Trioritise components'(node A03-325, figure 7.7) do not considerproduct structurein prioritising componentmanufacture. Figure 7.8 describes the situationwhereby all componentsare routed through profiling (nodesA03-332 or A03-333),dressing (node A03-334) and shotblasting (nodes A03-335 or A03-336)before 'Processprofiles' (nodeA03-338). Here modellingled to a recognitionof why material flow in downstreamprocesses was being largely controlled by profiling andwhy the UOB 'Dressprofiles' (nodeA03-334) was often a bottleneckand the causeof materialflow problems. 147 Process routes and operation details (tooling, settings, etc.) were based on operator knowledge and experience. Figures 7.8 and 7.10 showed that Process routes were relatively simpleand known by the staff involved but the information was not documented and could not be collated to provide the detail necessary for planning capacity requirements. Inventory levels for individual componentswere not consideredby shop floor supervisors when producing work to lists (activity A32, figure 7.6 and node A03-32, junction B, figure 7.7). One of the most important findings of the analysisis describedby figure 7.8 that shows the way in which shop floor monitoring is notified of profiled quantities. Profiling nests are based on standard practice and locally held data at the Plasma and Gas profiling machines(nodes A03-332 and A03-333, figure 7.8). Consequentlyplanning batch sizes for components does not take into account the occasions in which plate utilisation requirements force profiling operators to over-produce batch quantities of many components(and in somecases split batchesand under-producebatch sizes). Notification of variationsto batch sizesonly takes place after profiling, the designof the CEME centre must overcomethis problem. 9 Rework and WIP levels were not recorded in monitoring the shop floor activity (activity A3 4, figure 7.6 and UOB A03-3 37, figure 7.8). Customer haison was carried out at several levels in the company with the attendant problems of conflicting information being passedto customer's staff, this is shown in figure 7.3 (node Al. 1, A12 and A13). The model prompted a discussion concerning the need for drawing managementand production facilities (node A12, figure 7.3) with the outcome that this activity, as far as the CEME centre was concerned,was not necessaryand that drawings (hard-copy or graphics files) would be accepted directly from customers. In addition that Haison concerningdesigns would also not require a dedicatedfacility. Outcomesfrom the this stagedemonstrated the ability of MEFO-3 to provide a clear analysis of current operating practices. The following stepswere taken to fill someof the information gapsexposed by the analysis: 148 *A definitionof the productwas established,bills of materialfor eachIdt were produced andvalidated. Preliminary process routes for each component were designed to provide a basis for machine and capacity requirements planning. Subsequently the machine capacity requirementsbecame part of quotation specificationsfor equipmentvendors. BSD staff estimatedoperation and set up times in terms of machinesand labour hours. Operation details such as plasmaprofiling times, robot weld cycle times and press brake tool change times were designed in conjunction with potential machine suppliers and existing shop floor personnel. Many of the problemsmodelling exposed during the work were clearlyconcerned with the lack of documentedinformation, the control of manufacturingand particularlywith planning manufacturingcapacity and providing shop floor schedules.As a result, the project and subsequentfocus of the AS IS model shiftedto developa picture of runningthe CEME centreusing a computerbased approach for planningand controlling shop floor capacityand schedules. 7.6 THE DESIGN OF THE CEMIE CENTRE The managementand operationalprinciples of the CEME centre were modelled over a series of extendedsite visits and meetingswith senior BSD stafe over a two month period. The final agreedproposals for the centre are describedby the model 'Run CEME centre'. The scopeof the model is shown by the node tree diagram in figure 7.12 and node index in figure 7.13. By this stage in the project, orders were being placed for mechanicalhandling equipment, profiling equipmentand other plant on long delivery times. It was also consideredimportant, at that stage,that the developmentof managementand organisation of the centre took place in parallel. Consequentlyit was agreedthat the TO BE model would be used as the primary meansof establishingthe characteristicsof shopfloor control. The model was also intended to be used for presentationsto shop floor supervisorsand BSD group management. Mr J. Backhouse,Managing Director BSD Leeds Mr E Gibbs, Engineering Operationsmanager Mr S. Hilton, Project Co-ordinator. 149 AO I A-0 A2\ A3 All ýAA32 A12 A13 A31 Al A33 A21 A21 -7 A212 1-11, A22 A23 A213 ,p A221 A222 A00- A223 Aý24 FEO A03-223 Figure 7.12 Node Tree for IDEFO-3 model 'Run CEMIEcentre' 7.6.1 Phase 6 of the EDEFO-3methodology Model TO BE using IDEFO-3 The TO BE IDEFO-3 model comprisesten diagramsshown in the node index of figure 7.13. The purpose of the modelwas to describethe operationand informationrequirements for control of CEME centremanufacturing and the viewpointwas that of operationmanagement. The completemodel is shownin figures7.14 to 7.23. The key featuresof the new approach to kit manufactureusing the CEME centreare described by the TO BE model: The CEME centrewill operateas an independentmanufacturing unit constrainedby a unit businessstrategy (the details of the strategy was to be defined by BSD at board level at the time the model was produced) shown as a control in figure 7.14 and 7.15, node Al. The implication is that all componentsproduced outside the centre, by BSD or outside suppliers,would be treated as sub-contract(shown as an input to node AO in figure 7.14 150 and 7.15). Similarly all components,other than kits, manufacturedby the centrewould be treated as sub-contractcomponents (shown as an output from node AO in figure 7.14 and node A3, figure 7.15). Figure 7.15 showsthe key activities in the centre and indicatesthe dedicatedmanagement and planning for activity A3 - Manufacture kits. Costing and planning information from (node A2, figure 7.15) will be made availableto the centre managementactivity (node Al, figure 7.15). Costing and planning information (shown as a control on node Al, figure 7.15) consistsof aggregatecapacity and lead time information to allow realistic kit delivery datesto be discussedwith customers(shown as ('work an output - 'customer haison'). This together with feedback from work centres centre reports') will allow workable kit schedulesto be produced (an output from Al, figure 7.15). Work centre schedules,routings and process documentswill be issued by planning staff (outputs from activity A2 figure 7.15). In addition nesting information will be downloaded to profiling equipment and dispatch documents issued on kit completion (outputs from activity A2 figure 7.15). In the first stageof implementationthe centre will rely on documentsto record and control manufacturing. Software systemsfor the control of manufacturingwill have be able to support bar code readers and remote terminals as input and remote screensas output in addition to paper reports. This is not explicitly describedat this level in the model. Work centre reports (an output from activity A3 figure 7.15) will include batch manufacturingtimes, inspectioninformation and scrapand rework records. Shop floor (work centre) performanceinformation win be produced by activities A12 and A13 (figure 7.16). Liaison with customersand supplierswill be through the single activity (node Al 1, figure 7.16) and the staff involved will be the singlepoint of customercontact. e The activity 'Establish manufacturing schedules' (node A22, figure 7.17) provides schedulesfor the following outputs; Plate Profiling/nesting Work centres sub-contractmachining 151 Graphical component data is shown as an input to A22 (figure 7.17). Clear part recognition will be provided by component profile drawings in addition to part numbers on work centre routings and componentschedules. [AO]Run CEME centre [Al] ManageCentre [Al 1] UaiseWth customersand outside bodies [A12]Monitor machine performance [Al 3] Controland monitordirect staff [A14]Control and monitorfinance JA2]Plan manufacturing [A21]Carry out manufacturingplanning [A2111Maintain drawings [A212]Control Processinfo, toolinginfo, and NC data [A213]Establish consurnables requirement JA22]Establish manufacturing schedules [A221]Establish nesting data JA222]Produce schedules and documents [A223]Control vwrk centreschedules [A224]Track inventory [A03-221] Establish nesting data [A03-223] Control work centre schedules [FEOA22F] Establish manufacturingschedules JA23]Control Inventory [A3] ManufactureKits JA31]Make components and assemblies [A311] Produce and finish profiles [A312] Process profiles JA3121]Set up for batch [A3122] Run machining process [A3123] Close batch JA313]Configure and weld assemblies [A32] Inspect components and assemblies [A33] Palletise Kits for dispatch Figure 7.13 Node index for 'Run CEME centre' 152 Work centre capacities,BOMs, process routes and inventory levels will constrain shop floor schedules(activity A22, figure 7.17). Customer drawings will be maintained by NC programming staff, who are the primary users of contract customer drawings, to produce profiling instructions and robot welding, data and part profiles for processroute documents(nodes A211 and A212, figure 7.18). The generation of nesting data (activity A221, figure 7.19) and work centre schedules (activity A223, figure 7.19) will be an iterative process,shown by the feedback and feed- forward loops betweenthe two activities ('rough cut profiling schedules'and 'variations to batch size' respectively). Inventory will be tracked (activity A224, figure 7.19) using predicted levels, inventory counting and exceptions(real time feedbackfrom work centres). The diagram 'Establish Manufacturing Schedules'(node A22, figure 7.19) was central to the designof the CEME centre. Therefore, as part of the TO BE model, an FEO diagram (figure 7.21) of 'Establish Manufacturing Schedules' was used to introduce the CEME, centre proposalsto shop floor supervisorsand in preliminary discussionswith software vendors. The behaviourof two functions'Establish nesting data' and 'Control work centreschedules' (nodeA223 and nodeA221 respectively,figure 7.19)was critical to shopfloor control for the centreand both were decomposedto IIDEFO-3diagrams, figures 7.22 and 7.23. Figure 7.22 showsthe relationshipbetween the generationof nestinginformation and its influence on the control of work centreschedules, the three outputsfrom node A221 (figure 7.19) 'Plate requirement''Nesting information' and 'Variationsto batchsize' are correlatedwith three associated'release' UOBs in figure 7.22. The asynchronousOR junction J1 representsthe decisionnecessary to proceedto the oneor moreof the three 'release'UOBs and the 'Plan profile nests'go-to referent. The constrainton the decisioncan be correlated with the control on nodeA221 (figure 7.19) 'Profiling processinformation' which, when decomposedto J2, indicatesthat proceduresfor schedulevariation and nestingtogether with planning horizons will be necessary. The IDEFO-3 diagram also describesthe situationthat in order for 'Plan profile nests'to proceedrough cut profiling schedulesmust be availableas a result of 'Control work centreschedules' (node A223, figure 7.19). The approachadopted in the CEME centre will be to treat nesting schedulesas a constraint rather than allowing them to be the driver for work centre schedules,this is describedby the IDEFO-3model. 153 The IDEFO-3 diagram figure 7.23 shows how the function 'Control work centre,schedules' (node A223, figure 7.19) behaves. It shows the PFN initiated by a decision Ounction JI) to 'Run scheduling software' or 'Use existing schedule'. The constraint on the decision is 'Changes to schedule' a control correlated with figure 7.19 (node A223). The constraint on proceeding with either UOB 3 or 4 (accepting or rejecting the schedule)is the need to balance routine or exceptional capacity with component demand and due dates. Work centre capacities and 'Component level detail Kit schedules' are the controls correlated with figure 7.19 (node A223). A schedulecan be tentative (to assessthe implications of the component nests it creates) leading to the UOB - 'Release long term profiling schedule', or firm in which case the UOBs 'Establish predicted inventory levels' and 'Releaseshort term work centre schedule' follow. The outputs from these three UOBs can be correlated with outputs from node A223 (figure 7.19). The IIDEFO-3 diagram reflects the need to use rough cut profiling schedulesas a constraint on 'Establishing nesting schedules' shown in figure 7.19 (node A221). If a scheduleis rejected (UOB 3, figure 7.23) the decision concerningwhich UOB follows is constrainedby a balancebetween capacity, processroute options and a detailed component schedule. The situation is described(in figure 7.23) by the relationship betweenjunction JS and UOBs 8,9,10 and 11 and the correlation of the controls 'Component level detail kit schedule',work centre capacitiesand processroute options. 154 F-e 0 9D CL a E 20 0 E0 at W 0 0 0 in E 0 CL C0 rL ow Iý9 I_ (.. 0 Dow EE o" C) ( Ou,oc .0 C .9C c C) 0 U w J= (L) rE C-)w 0 (U CL e- (L) C .0Em Co 00E W 2 0w E (D 4) 0. Im 0%- m E Eo (D cr CL 0o 0 9 0 (D Figure 7.14 Node A-0 Run CEME centre 155 ()0 13CL e CL WC a) a) E' "2 A0- E -m =m 0 :3 0E E i5C> U)AL 0 :3 M 42c: 0 < CLl j Loio CL ti) 0 E0 92.r- od 0) *5 ý ý x 0 e8 12 ia 0 0 1 Ll ZE :2 -0 0 EL-9 z930 0 U M 1 0)(0 Eo C, 0 7 - e J r_ -Z> r- 0 eg0 c> 0 %0 lt L t400 fi 0 00 E Cm -- A 1 0. 0 > 75 w 4ý, E 00 m u -bl -0 ý 5 r- 4) E O 2 CL) 0 2E 0 tu006 CL m 0 m (U 0 2- Z gij m0 U) t) im gl. E-L 0 lz 1 Figure 7.15 Node AO Run CEME centre 156 11 c0 (1) (1) :p (1) en E f2 2 t) 11 75 c (D cc r_0) gl; zj 50 0m a) to 0 9E in >, = :3 -iü CoIM 12- im. a. u) r6 N 04 GA 0 u 1 I l 0 0 0 0 Co 0. 0 u m 1 0 92 1 0) Iti 0 CL 92. a- 0 a c tu (4) aN Lor- Z0 150 w 0 00 0 CL w2 CD r_ c3) 0 0 12. m r_ ch c» M Co g 1 r- 0 )0CL L w- V) 9 L) 424 - ic . -- ý e In :0 10 «0M A 0 :3 M 1:ý - c3) ja 0) m :i te.. t2 Q0 O 4) 0 Z to ý0 r_ zm 4) a) E < 0 -0 U (L) <= EIm Figure 7.16 Node Al Manage centre 157 c j= 0 to0) r_ m c 0) * «a c . (0) 0 U) (L) M ZZ r- ý r_ t n. 0 (0 (P t) - - 0 E M 5 Z0r I 0 0m«a , Z UM 0 0 c2 Z P- 0 u) Co 40 U). Q £2 ik L ÄL cn0 A Q :E -0 0. C, wE ö 0) (4) . 2 m 9% Im U-(0 m 8! 0 > .02 ?'c(0 0E 1 = u m c 0. LZ rL >>x E E U) C> Z,v -Zr4 C.) I .ýI 15 90to 0 m12. IE :: fß ,n - 22 00 :6 -0 - .0 _j 0 0 I -Z Figure 7.17 Node A2 Plan manufacturing 158 11 9 ý:C, s?2 EE CL) £X m 0 Qý 0 92. ig tu 20 U.: 0. CL '13 ý 2 2 E r- ci 0 :Z 2! r_ 2 cm Co r_ < JL.0 M r_ 10 0 r_ 0 t!. c:ý 0, Z CL 0-2 111F -L ce j u 1-.&; >.. 9 , w - :08 2 WU u. 2. 5Z re 0 0 a)_j o 0 92. 00 0 C» 12 5 .2 vli L 2 mCL E 0 u2 CL Figure 7.18 Node A21 Carry out manufacturingplanning 159 13 N '*' E .I b- ýM 00 :p0 32 to ýM 0 m 1 0-X r_ M0 4) - (1) (0 mm e 0w c» fj > j9 8 Z 4 Z0 -c cy od c ý0- 42 :E- 32 m J, 0 0) -Ä L.E 0 a% w cn 02E 'm A F.: 0 UM -2r- 2' r- Z0 ig m0 -0 >x E. 2 A -, . e- 0 (D (D 0 c4 1 (L)A c%Iml pgl w« -E (0CL --ýt 10 c4 E cm Zm im 0 Ch 2 N 0 M 7. Co > JD to 9 cj w 00 U) CL U EU) c: ý 0 lte<"! E 0 Figure 7.19 Node A22 Establishmanufacturing schedules 160 U v t- 1 r_ p4 0 cx CC 00 4D 0 Ca. 0 w a. 0 CL je 19 II W6in r4w mX Z - E JO- 22 c V 2CL < CL 0 j2 u e0 0 mý2 r_ < A& .2 0. - C" E jg02 v;00 0 gý l! A Z i -Z m r. , 0 ýI, 00 r 0-t0 . CL A49 9ý U) E ,Z .; od- :50r_ m00 im. 0 u) E 0)E0 0. - in E u Ei = 2.1 O 0 ED 6 f. 0 ------.-«rn c: < .20- Z13) 00 m 0M 4) I!Rý 41 A IL (X :3 0) r- Figure 7.20 Node A3 Manufacture Kits 161 U) N M A CL 0 x0 0 c; w -E 0 im od 0 0 44 > E EE mE A A t= C) Ir :y 12 e 0) C IM 3, r- -0 2! 0 in. o cýw Xa - ,- 0) c - *E 0m «E«e 2 2 ?00 22 8v -1- m e 75 e >0) CL0 2E-,m0 0) .-m j .to CJ im. oZ IU92 ED 00 ,'to . ge ý- uaA 2 A = A 0 c: 75 J-3 a u) . j 4) c9 r m :3 KM in p4 90 r- c3) r- C) Z 0 0 _A j3. C) ui Q > (4 (0 to CD 0() (D 22 fl. 0% ; ý« (U C) 00 C -0 41 ci . cn cnw U. S cli m z» ci. Z 422 rA Z 2 A Figure 7.21 Node FEO A22F Establishmanufacturing schedules 162 A = cm E r- 0 oý 1 .!Z) Zý f-- 42 a0'. 0 rr (0 a) ur IU 0r- blý 42 0) cm a: 2 JO-ce 00- 0 -C» Z. JE 16a m t4 E r- %P iE 000000M M Co (p r- *r 3.0 Ja Cy da cä. > 12 CL 0 1-- C 020 *.0 0 0) 0 .-U 12p .'29 c,. 0. m 2 -r- 0M1E r_ CL V- j--- u-- c) r. N- 1 E 2510 U) CLeo_ 4.- C» 2 r_ JM 1 m.C: E «a A 3: 'm 012mr40m m :g0 :E -Ir Figure 7.22 Node A03-221 Estabfishnesting data 163 A E 4) 2s 0 oam 0. Eli . I I ,a 41 1 i -s CL «2 1. I lz CL 9,2. , 22to C 10 u0 .C u UN 0 1fit .291 1 CL < Ln mý C) 0) '9 1 -5 9 ts ' ? Cl-9 1E I v ii z -5, -I cccö ". N to C C) 00 xu (n. 0 to 1 Z:5 fJ 0(0a Figure 7.23 Node A03-223 Control work centre schedules 164 7.7 CONCLUSIONS The final phase of the 11DEF0-3methodology involves implementation of the changes describedby the TO BE model. Significant changestook place in the managementof the BSD Leeds sites during 1995 and a decision to build the CEME centre at another location was taken in mid-1995. This was due in some measureto local objections to the additional noise and traffic that would result from three shift operation. Consequentlysystem start up has been delayed. Implementation of many of the changes described by the model are currently underway (January 1996) The most significant is acceptanceof the operating approachdescribed by the TO BE model and the purchaseof a shop floor control system(in November 1995) configuredto support the model structure and definedin the BSD document (which is basedon the TO BE model) shown in appendix 11. The work of chapterseven has been analysed from two perspectives: (i) The casestudy site. Through the expansionof Spot and Contract manufacturingboth the manufacturing staff at TIF and the BSD project team had considerableexperience of plate handling plant and profiling and cutting equipment but limited experience of structured approaches to manufacturingcontrol and of computer based approachesto shop floor control. The initial proposal to use modelling as a meansof describing organisation,behaviour and information was treated with some scepticism. However as the project developed the AS IS model exposed the complexity of the existing situation and the design task ahead. Building the models, involving BSD staff as readers, and the discussionsthe diagrams provoked were recognisedas significant to the successof the project. Two formal presentationsof the case study findings and proposalswere made by the author to shop floor supervisorsand senior management at Leeds and one presentation was made to group management at BSD Stourbridge. SubsequentlyBSD adopted the proposalsdescribed by the TO BE model and incorporated them into their specification for shop floor managementand scheduling. The companyspecification based on the IDEFO-3 model is shown in appendix 11. It relies on the IDEFO-3 model diagrams (fisted in the final paragraph of appendix 11) figures 7.19,7.2 1, 7.22,7.23, ) and uses figure 7.19 to describethree key activities necessaryfor shop floor control (defined in paragraph I of appendix 11). In addition a BSD produced overview diagram of 'CEMIE Scheduling'was used in the specification(appendix II sheet5). 165 (H) The applicationand developmentof the IDEFO-3 approach. Application of IDEFO-3 relies on the proven modelling methods,IDEFo and IDEF3. It is the integration of the methodsand the synergyensuing from integration that is being tested by the case study. It was evident during modelling that the interface between the two methods revealedaspects of systembehaviour in describinghow IDEFo controls influenced processes and decisionsand how IDEFo outputs were produced. The IDEFO-3 TO BE model, 'Run CEME centre', proved capableof incorporating and describing the recommendationsfrom the AS IS model discussedin section 7.5.3. Using IDEFO-3 in the TO BE application, where diagram detail cannot be checkedagainst e)dstingsystems, revealed its ability to validate the model by ensuring ICOMs in parent IDEFo diagrams and PFN structure correlate after decomposition. It was found that all IDEFo controls are not necessarilydirectly related to decisions or Junctions. In such casescontrols are linked to precedencearrows to indicate which aspectof processflow logic they constrain. An example is shown in the precedencelogic preceding UOB I in figure 7.23 where the controls TOMs, process routes', 'Inventory levels' and 'Variations to batch size' are necessaryfor the UOB 'Run schedulingsoftware' but do not influencethe decisiontaken at the precedingJunction J1 or the alternativeUOB 'Use existing schedule'. The stage in modelling that an DDEF0function becomesa scenario for a PFN is key to the IDEFO-3 approach. For AS IS models it has proved to be, in the case studies to date, the stagein modelling where descriptionsby domain expertsrely on a logical narrative of events, i. e. a sequencerather than a descriptionof aggregatedfunctions. In an D)EFo diagram, at this stage,a description can becomea 'waterfall' of functions indicating that functional modelling is turning into processmodelling. In TO BE models the stage at which the logic of events is necessaryto understandbehaviour is the stageat which the interface between I1DEF0ICOMs andPFNs must be described. Decomposition of a UOB in an IDEFO-3 diagram was found to be necessaryin one case shown in figure 7.9. Further decompositionof UOBs could, if necessary,be accomplishedas in conventionalIDEF3 modelling. Elaboration documentsfor UOBs and junctions were not found to necessarybecause the correlation of IDEFo ICOMs and PFNs gave the information necessaryfor model analysisand systemdesign. This case study gave an opportunity to refine the EDEFO-3methodology user guide and the final version is presentedin appendix2. 166 CHAPTER 8 DISCUSSION, CONCLUSIONS AND FURTHER WORK 8.1 MODELLING IN NLANUFACTURING ENTERPRISES The aim of this researchwork has been to further the use of graphical modelling methods as a meansof gaining insight, designingor improving performancein manufacturingenterprise systems. During its course it became evident that no single modelling method would be sufficient to capture the complexity of a manufacturingenterprise and that the selection and use of a method is subjectto the purpose of the modelling exercise,the nature of the system being investigatedand experienceof the analyst. For manufacturingenterprises few methods have emerged to be generally accepted. To model the interaction of facilities, people, information and material is often so complex that a specifically designedmodelling method, such as IDEFo, is required. Although it only usesa simplegraphical box and arrow construct with minimal model syntax its ability to model complexity is remarkable. Experience is necessary to build an effective model however, when used to communicate system descriptions,untrained users can quickly understandand interpret diagrams. It is evidentthat continuousimprovement and systemchange and developmentis the norm. The researchconcludes that the IDEF modellingmethods significantly contribute to structure and guide the changeprocess. IDEFo has advantagesas a manufacturingenterprise modellingmethod in that it is in the public domain,it is widely understood,is well supported 167 by software, and it has a proven record of successand has been adopted as a standardin US defencemanufacturing. Although IDEFo has been used extensively by the US defence industry its use outside is limited. The review of IDEF methods together with the survey in section 3.5 indicate that for UK manufacturingenterprises modelling is largely led by academicsor consultantsand that many aspectsof manufacturing system design evolve from establishedpractice. For many enterprisessystem models consist of disparatemethods and diagramssuch as discrete event simulation,CPA, DFDs, flowcharts and companyspecific diagrams. The graphicalrepresentation used in a model is an expressionof knowledge about the system under review, building exhaustiveIIDEF models consumestime and staff resourcesresulting in decisionsbeing made before a model can be completed, in addition systemscan change during modelling and validation. A pragmatic view is that in today's competitive, manufacturingenvironment it is unrealistic to expect UK manufacturingenterprises to have the time for protracted modelling programmesas a precursor to change or system design. This can be countered by using IDEF methods in a rapid, rough-cut way, diagrams can be produced quickly, in manual form if necessary,and the samemodel syntax can be used at a strategic or operational detail level. In this way it can be used just as effectively as for exampleCPA and in due course,become as universallyaccepted. The use of a genericor referencemodel was demonstratedin section5.3 this approach significantlyspeeded the modellingprocess but of course relies on sufficient reference modelsof the elementsof manufacturingsystems being available in the publicdomain. To be fully effectivea standardmethodology, CASE tool usingstandard data formats and a library of modelswould be necessary.Other engineeringdisciplines have extended the practiceof using the standardlibrary approachroutinely (suchas NAG library routinesand retrieval (variant)process planning systems). It seemsunlikely that extensiveuse will be madeof referencemodels but sincethis aspectof the researchwas first published(Colquhoun and Baines,1991) several other authors have advocated the useof referencemodels. 168 It is evidentfrom this researchthat there is seldoma clear distinctionbetween different modellingmethods and there is little clear guidanceor commonunderstanding on what aspectof an enterpriseparticular methods are best applied. The lack of conventionsand commonpractice for enterprise,system, or processdesign is in sharpcontrast to product descriptionand designwhere the rangeof modellingformats is well defined(for example graphicalproduct descriptions are used for kinematicmodels, finite elementmodels and tool path models)and the roles, scopeand interfacesfor the various modelsare commonly accepted.The manufacturingenterprise framework detailed in section5.4 hascontributed to a rationalperspective of modellingmethods for manufacturingmethods and can contribute to their routineuse. 8.2 APPARENT LMMATIONS OF USING IIDEFo The focus of the IDEFo method is on activities that transform inputs into outputs under the influenceof their controls and mechanisms.Its limitations he in its limited ability to describe the effects of controls and the timing of system behaviour. It is capable of depicting the infonnation or objects necessaryfor control. The content and nature of information can be described using labels, however it cannot be used to describe how controls can affect functions or give information on their use. It does not attempt to describe the logical or consequentialaspects of systemsor the interaction of controls with activities or processes. Much of the criticism levelled at IDEFo is related to its inability to representtime. However research suggests that there are three viewpoints of 'time' in manufacturing systems modeffingterms. 1. The first is the passiverepresentation where time is describedby a label attachedto a graphicalmodelling symbol as in CPA or the GRAI grid. 2. The secondis where time is an integral part of the graphical representationas in discrete event simulation, where the graphical description can change over time, where, over a 169 time period, events (that changethe state of a system)are predicted and where waiting, queuing and processingdurations are considered. 3. The third is a more relaxed viewpoint (in modelling terms) where a modelling method representsthe temporal aspectsof a systembehaviour and can capture conceptssuch as; after,before, during, when and until. Fundamentalto all the aboveviewpoints is the needto understandand describe the sequence of eventsprocesses or activities. Howeverconsideration of sequencecan act as a constraint in isolatingand understanding functional relationships. One of the strengthsof IDEFo is that an analystcan in fact ignore 'time' and sequenceand concentrateon aggregatedactivities. Criticsof IDEFooverlook this positiveaspect of the method. In the secondviewpoint it could be arguedthat the clarity of an IDEFodiagram would be lost if it attemptedto modeltime, as manysystem aspects neither change over time or havea time relatedsignificance. In any event simulationis well servedby an extensivebody of knowledge,the key to providinga time aspectfor IIDEFOmodels is best accomplishedby providing integration paths betweenan IDEFo descriptionand existing approachesto simulationand IIDEFO-3 satisfies this criteria. The DDEFO-3approach incorporates the best of IDEFo and IIDEF3 using IIDEFoat a higher level to gather aggregated functional descriptions and then using those functional descriptions as a scenario for IIDEF3 PFNs where the sequenceand temporal aspects of systembehaviour can be described. In the surveyreported in section3.5 andthe researchreported in section3.3 it is evidentthat DFDs are perceivedto have a similar role to IDEFo and that they are used to model manufacturingsystems. The perceptionis perhapsbecause DFD approachesare more widely usedin softwaredesign and are thereforeconsidered more matureand as a result moremodelling practitioners have used DFDs. IDEFohas several advantages over DFDs for manufacturingenterprise applications. The structureof a DFD hasno formallayout making it difficult to seewhere to start readinga diagram. The box usedby IIDEFoto represent 170 functions provides four faces to structure connecting arrows using the formal syntax. The circles used by DFDs do not provide this feature and a confusing crossing arrow structure can be the result, feedback arrows for example can be difficult to recognise when no positional structure exists for the bubblesused to representprocesses. No limits are set on the number of processesused in a single DFD and a large number of processeson a single diagram can be difficult to understand. In addition the file and data store symbolsused in DFDs are not necessarilyan important modelling characteristicfor manufacturingenterprise modelling. 8.3 THE CASE FOR THE EDEFO-3METHOD As a modelling method for manufacturingenterprises IDEFO-3 extends IDEFo and enables the representationof additional manufacturingcharacteristics in a single model. IDEFO-3 is able to represent features of a system not representedby either IDEFo and IDEF3 as individual models. More precisemeaning can be given to controls by assigningthe controls identified as necessaryfor IIDEFOfunctions, to stagesin IIDEF3PFNs. Consistencychecking and model validation is also inherently achieved through the need to allocate inputs and outputs preciselyto a PFN structure. In modelling a systemthe first step is to capture its significant activities, systemelements and important information. Systemdescription becomesmore complex at an operational level where the necessityto understandbehaviour is essential. In practice detailed descriptions given by domain experts rely on describing a logical flow of processes or a flow of information rather than on abstracted descriptions to describe operational detail, this re- enforcesthe use of a sequencedescription such as a PFN. The IDEFO-3 notation provides a time line for IDEFo functions and permits the creation of a single model that is capableof describinga greater range and depth of manufacturingsystem behaviour. The IDEFO-3 approachextends modelling capability by describing: * The influenceof infonnation or objects on processes. 9 The controlsand conditions for activitiesto initiate,stop or proceed. 171 * The controls necessaryfor decisionsto take place. * The mechanisms(stafT, facilities or systems)responsible for decidingon alternative processsequences and for carrying out processes. * The consequencesfor objects,infonnation or dataprocessed on the sequencesof processes. In this way IDEFO-3 gives a means of gaining insight, designing and improving the performanceof a manufacturingenterprise system hitherto not possible. The method (appendix 2 contains the user guide) was validated through its trial in a complex, five case study (reported in chapter 7) that tested its syntax, methodology and modelling capabifity. A total of six days, over a two month period, were necessaryat the casestudy companyto gatherinformation for the modeland a fimher fifteen daysto analyse, validateand presentthe modelto the company. The researchdoes not draw any conclusions over the time involvedfor the new approachuntH more appEcationsare carriedout. To build the AS-IS model it was first necessaryto collect relevant information from the various domainexperts then to reconcilethe informationinto a coherentpicture of the way the manufacturing system operated. Prelin-dnarymodeHing starts with an incomplete understandingof the overaUsystem, at this stageIDEFo is particularly good at forcing an analystto identify the important features,both activitiesand information,using a top-down approachto describeinformation in a functionalcontext. In the casestudy none of the staff involved had previousknowledge of IDEF model1ing.Despite this diagramvalidation was accomplishedwith few ambiguitiesand inconsistenciesand the terminologyused was quickly resolvedwithout the needfor a formal glossary. IDEFo functionsbecome scenarios for a PFNs at a key stagein the IDEFO-3approach. In the case study it was clear when the knowledge of domain experts was expressedas a sequentialdescription of processes,activities or eventsusing their relative positionsin the sequenceas a frameworkfor discussion. 172 At this level of detail in conventionalIDEFo modelling an analystwould normally have to interpretthe systemdescription and extract a functionalview. The transition from an IIDEFOfunctional diagram to an IIDEFO-3 diagram provides a behaviouraldescription using the correlation of ICOMs and PFNs- In the case study this madeit possibleto understandhow manufacturingwas carried out in the AS IS model and how a softwaresystem could be configuredto support shopfloor schedulingand operationsin the CEME centre (TO BE) model. It was also found that elaboration documentsfor junctions and finks were not necessarybecause the information provided by correlation providedthe appropriateinformation. The casestudy demonstratedthat IDEFO-3 was capableof capturing and representingthe perceptionsof domain experts. The IDEFO-3 approachsignificantly helped the analysisof the AS IS model,to identify differencesbetween their perceptionsand resolvethe model into a commonly agreed system description. The AS IS had sufficient resolution to allow meaningfulanalysis and providedthe basisfor developingthe TO BE model. The indications are that the model minimiseddistortion of the systemdescription to fit the characteristicsof the modellingsyntax. The arrow structureflowing from left to right in a PFN, is intuitive helping IDEFO-3model diagramsto be usedas the basisof communicationbetween the systemparticipants and the analystwithout the needfor an extensiveknowledge of the modeflingmethod. It has been evident in all the modeffing undertakenthroughout the researchthat it is important to minimisethe numberof diagramsfor a model to be acceptedand deemedcredible by non- modellingstaff. The IDEFO-3approach does minimise model size. 8.4 FURTHER WORK This researchis beingextended in two areas: 173 (i) Using EPSRC-CDPfunding (grant ReferenceGR/K39400) the researchdescribed in chapter three and four is being extended to distil current best practice approachesin modellingmethods for manufacturingenterprises. The researchaims to: " Establishthe extent of the UK user basefor graphicalmodelling methodsapplied to the analysisand designof manufacturingsystems. " Categorisethe type and natureof applicationand industry andto capturebest practice. " Report on the effectiveness,benefits, resource implications, application time span and training necessary. " Disseminatefindings and producea video and manualfor exploiting the methodsto their M potential. The first phase of the project, started in 1995, was a national survey of manufacturing companiesto primarily identify a group of manufacturersusing modelling methodsand to provide an indication of which methodsare being used, the types of manufacturingsystem being modelled,the characteristicsof industrial users, the barriers to use and to establish measuresof successfor application. Following the survey a series of longitudinal and snapshotcase studies will be usedto monitor the way that UK manufacturersuse modelling methods in the changeprocess in their organisation. In parallel similar case studies in companiesnot using modellingmethods will provide a basisfor comparison.The final phase of the work will disseminatethe findings through a manual and video based on a methodology,application guide and best practicecase studies. Preliminaryresults are being published(Small, Baines, and Colquhoun, 1996, Baines,Small, and Colquhoun,1996). (H) It is necessaryto build a numberof modelsusing the IIDEFO-3method. To this end, again using EPSRC-CDPfunding, another case study site is being consideredwhere the methodwill be usedin an automotivebody blankingplant currentlyunder construction. The plant has no establishedprocedures or working practicesand the method wiU be used by personnelhaving no previousinvolvement in its development. In more generalterms the following further work hasbeen identified or has arisenfrom this research: 174 The developmentof a software tool to support IDEFO-3, to reduce the time involved in diagramproduction, to provide syntaxchecking and for the methodto be more acceptableto industry. integration with existingIDEFo and IDEF3 softwaretools would be a requirement to allow import and "port of existing models and to be compatible with modelling standardisaflon(Cauthom, 1993). The author / readervalidation, control and managementof models generallyhas not been extensivelyresearched yet this aspectof modellingwas found to be of critical importance. For manufacturersto gain confidencein the validity and effectivenessof models and to contribute to the successof manufacturingenterprises, model validation will need fiu-ther attention. Similarlylittle attentionis given to the analysisof completemodels. No evidencewas found of structured methodologiesbeing extended to incorporate analysis and interpretation, phaseswhich often rely on valuejudgements being made. The structurednature of fnished models has the potential to lend itself to the developmentof a set of generic rules or guidel.ines related to the characteristicsof diagramsthat could flag potentialsystem problems or solutions. 175 REFERENCES Ackoff, K L., 1959,Games, Decisions and Organisations,General Systems, IV, 145-150. Ackoff, K L., 1962,Scientific methoc4optimising applied researchdecisions, 109 - 175, (JohnWiley & Sons,New York, NY). Agresti, W. W., 1986,New Paradigmsfor softwaredevelopment, (= ComputerSociety Press,Washington, DC). Aguiare,M. W. C., and Weston,P, H., 1994,SEW-OSA - SystemsEngineering Workbench ccntredon CIM-OSA, Proceedingsof the 101hInternational conferenceon ComputerA idedManufacturing Rohoticsmid Factories of the Future, CARS- FOF/94,Canada. Albrecht, K., Jensen,K., and Shapiro,R., 1992,DesignlCPN, A toolpackagesupporting the useof colouredPetri nets,(Meta Software,Cambridge, MA, USA). Alting, L., 1986,Integration of engineeringfunctions/disciplines in CIM. Annals of the CIRP, 35 (1), 317 320. Alting, L., andZhang, H., 1989,Computer aided process planning: the state-of-the-art survey.International Journal ofProduction Research,27 (4), 553-558. Anderton,R., 1989,The needfor formal developmentof the VSK Yhe ViahleSystems Modek Interpretationsmid applicationsof StaffordBeer's VSMed. R_Espejo & R. Harde,39-51, (Wiley, Chichester). Ang, C. L., and Gay, I;L K. L., 1993,IDEFO modelling for project risk assessment, Computersin Industry, 22,31-34. Ashby,K W., 1973,Some peculiarities of complexsystems, Cyhernetic Medicine, 9,1-7. AT&T ISTEL, 1995,Generateyour 97TNESS simulation models automatically with ProSim,Marketing brochure, (AT&T ISTEL, Redditch,Worcs. ). AUGRAI, 1987,Programme of the I st AUGRAI event,September, 1987, (The Association of usersof the GRAI method,The Universityof Bordeaux). Bachert,P, F., Evers,K. H., andRolek, E. P., 1983,IDEFO/SAINT SAM Simulation: HardwarefflumanSubmodels, Proceedings of the 1983IEEE National Aerospace andElectronics Conference,NEACON, 1066-1070,(IEEE, Piscataway,NY). 176 Baines, P, W., 1984,Production control assessmentcriteria in batchmanufacture, PhD Thesis,UMIST. Baines,P, W., andBetts, D. J., 1984,Avoiding the pitfalls of computerisation,Yhe British Journal ofAdministrative Management,34 (3), 87-99. Baines,P, W., and Colquhoun,G. J., 1990,IDEFO an Integrationand Design Tool for Engineers,Assembly Automation, 10 (3), 141-145. Baines,P, W., and Colquhoun,G. J., 1991,The useof a GenericIDEFo model to capture processplanning functional information, Integration and Management of Technologyfor Manufacturing,1, EEMwiagement of TechnologySeries 11,371- 378, (PeterPeregrinus, London). Baines,P, W, Small,C., Colquhoun,G. J., 1996,Modelling methodsin the manufacturing changeprocess, 7he 12thNational Manufacturing ResearchConference, (NMCR) September1996, The University of Bath, (awaitingreferees reports). Banedee,S. K, and A]-Mgik-i, 1., 1988,A structuredapproach to FMS modelling, International Journal of ComputerIntegratedManq/4cluring, 1 (2), 77-88. Barton, R. PL, 1994,Implementing a relationaldatabase for an accelerated-life-testfacility, IEEE Transactionson Reliability,43 (1), 11-21. Bauer,A., Bowden,J., Browne,J., Duggan,J., andLyons, G., 1994,Shopfloor control systems,(Chapman & Hall, London). Beer, S., 1979,7heheart of enterprise,(Wiley, Chichester). Beer, S., 1981, Brain of thefirm, 2nd ed., (Wiley, Chichester). Beer, S., 1984,The Viable SystemsModel: its provenance,development, methodology and pathology,Journal of the OperationalResearch Society, 3 5,7-26. Benjamin,P. C., Mayer, P, J., andBlinn, T. M., 1992,Intelligent supportfor simulation modeldesign from IDEF3 descriptions,Procee&ngs of the Conference'Business processdesign -A strateVfor corporateinformation mawgement', October, (IDEF UsersGroup, Kettering, Ohio). Benjamin, P. C., Mayer, R. J., andGraul, M., 1993,An integratedapproach to cost benefit analysis,Proceedings of the conferenceIDEFwid the corporateinfrasiructure, May, 174- 180,(IDEF UsersGroup, Kettering,Ohio). Benjamin, P. C., Cullinane,T. P., and Mayer, R. J., 1994,The role of IDEFOand IDEF3 in BusinessProcess Improvement, Proceedings of the conference,Reinventing the Technologies Richmond, enterprise of change, Virginia, May, 158-170, (IDEF UsersGroup, Kettering,Ohio). Bertalanfly,L. Von., 1956,General systems theory, GeneralSystems, 14 1- 10. 177 Bertalanfly, L. Von., 1962,General systems theory -A critical review, GeneralSystems, 114 1-20. Boardman,J., 1990,Systems engineering, wi introduction, (PrenticeHall). Bok, S. W., andNee, A. Y. C., 1988,NHCAPP A micro computerbased process planning system,Journal ofMelalý-orking Technolqjy, 17,21-3 1. Bond4 k B., andPritsker, A. A. B., 1985,Alternative analysesfor dynamicsmodels, Proceedingsof the Europeati Conferenceon Al Applied to Simulation,Simulation series, 18 (1), 324 1, (Societyfor ComputerSimulation, San Diego, CA). Booch, G., 1991, Object Orienteddesign with applications, (BenjaminCummings, Menlo Park, CA). Bowden, K, andBrowne, J., 1991,An approachto the productionenvironment design task within factory co-ordination,Computer IntegratedMaiwfacturing Systems,4 (1), 42-50. Britton, C., and Doake,J., 1993,Soft%, are systemdevelopment, 109-127, (McGraw-]HWL Maidenhead). Brovoco, R. R., 1991,A lifetime of lessonslearned in using and developingstructured methods,Proceedings of the conference', 4dvancedmodelling methods teclmoloSy,'Automation androbotics Research Institute, Fort Worth, Texas, 10-28. (IDEF UsersGroup, Kettering, Ohio). Bruce, T. A., 1992,Designing quality databaseswith IDEFIx information models,(Dorset House,NY). Buchel,A.. Breuil, D., andDoumeingts, G., 1984,Comparison of DesignMethodologies, Characteristicsand Deficiencies, Production Management Systems: Strategies and Toolsfor Design, 17-31, (Elsevier,North Holland). Busby,J. S., andWilliams, G. M., 1993,The valueand limitations of using processmodels to describethe manufacturingorganisation, 7he International Journal of ProductionResearch, 11 (9), 2179-2194. CAM-1,1980, Compositeview ofmanu/bauring architecture,Report no. DR-80-ATPC-02, (CAM-1, Arlington, Texas). Cauthorn,J. E., 1993,An IDEF modelrepository, Proceedings ofIDF-F and the corporate infrastructure,University of Maryland, 181-191, (IDEF UsersGroup, Kettering, Ohio). Cavaleri,S., and Obloj, K., 1993,Management systems, a globalperspective,87-92, (Wadsworthpublishing, Belmont, Ca. ). 178 Cesarone,J., and DobrzenieckiýA., 1992,IDEF modeflingin the next generationof manufacturingsystem, Proceedings of the EnterpriseIntegration Strategies Conference,Da qviona Beach, May, (IDEF User Group, Kettering, Ohio). Chadha,B., Fulton, P, E., and Calhoun,I C., 1990,Case study approachfor information integrationof materialhandling, Procee&ngs of the 1990ASAIE conference on computersin engineering,19-27, (ASME). Chadha,B, Fulton, P, E., and Calhoun,I C., 1994,Design and implementationof a CIM information system,Digineering with Computers,10,1 -11, (Springer-Verlag, London). Chang,T. C., andWysk, P, A., 1985,An Introduction to AutomatedProcessPlanning Systems,(Prentice HA EnglewoodCliffs, NJ). Checkland,P. B., 1981, Systemsthinking systemspractice,(Wiley, NY). Checkland,P. B., 1989,An appficationof Soft SystemsMethodology in action,Rational analysisin aproblematic world, ed. I Rosenhead,101-119, (Wiley, NY). Checkland,P., and Scholes,J., 1990,Soft SystemsMelhodolqy in Action, (Wdey,NY). Chen,P. P-S., (1976), The entity-relationshipmodel - Toward a unified view of data,A CM Transactionson DatabaseSystems, 1 (1), 9-36. Chu, Chso-Hsien.,and Wang,Hsu-Pin., 1988, The useof artificial intelligencein process planning,InteniationalJournal of OperationsandProduction Management,8 (1), 4-16. Coad,P., and Yourdon, E., 1990,Object-orientated analysis, (Prentice-Hall, Englewood Cliffs, NJ). Coad,P., and Yourdon, E., 1991, Paradigmchange, 7he Coad k1ter. Newadýcznces in Object-Orienled,4nalysis andDesign, (ObjectInternational Inc., Austin, Texas). Codd,E. F., 1970,A relationalmodel of datafor large shareddata bases, Communications of the.4 CM, 13 (6), 377-387. Colquhoun,G. J., 1988,An examinationof processplanning in a CIM environment,AfSc Dissertation,(Liverpool Polytechnic). Colquhoun, G. J., and Baines,K W., 1991,A GenericIDEFO Model of ProcessPlanning, International Journal ofProduction Research,29 (11), 2239-225 7. Colquhoun, G. J., Baines,R. W., and Crossley,K, 1993,A stateof the art review of IDEFo, International Journal of ComputerIntegratedManufacturing, 6 (4), 252-264. 179 Colquhoun,G. J., and Baines,R. W., 1993a,IDEFo-A survey of use and researchin the UK and Europe, Proceedingsof the 1993 IDF-F User Group Conference- Global Competitiveness,Integrating Process,Information and Technology,October 1993, Salt Lake City, Utah, 13 - 28. (IDEF Users Group, Kettering, Ohio). Colquhoun, G. J., and Baines,K W., 1993b,A comparativeinvestigation of manufacturing systemsanalysis techniques, Proceedings of the Tenth Conferenceof the Irish Manufacturing Committee,492-502, University College Galway, (Officina Typografica, Galway University Press). Colquhoun, G. J., and Baines,R. W., 1994, A 'Processdescription' approachto manufacturingsystems modelling, Advancesin Manufacturing Technology VIIII, Proceedingsof the Tenth Conferenceon Manufacturing Research, Loughborough University of Technology, 360-365, (Taylor and Francis). Colquhoun, G. J., Baines,R. W., and Crossley,R., 1996, A compositebehavioural approachto manufacturingsystems modelling, International Journal of Computer IntegratedManufacturing, Accepted by referees- awaiting publication date. Colquhoun,G. J., Gamble,J. D., and Baines,R- W., 1989, The Use of IDEFo to Link Design and Manufacture in a CIM Environment,International Journal of Operations and Production Management,9 (4), 48-65. Combefic,D., 1978,Experience with SADT, Proceedingsof the National Computer Conference(AFIPS) 'User experiencewith new Softwaremethods'631-633. Crimp, M., 1981, TheMarket Researchprocess, 7 1, (PrenticeHall International, Englewood Cliffs, NJ). Crossley,T. R., Davies, B. J., and Richardson,B. D., 1978,Progress Report, Preliminary Design of a Job Shop Control System,I January 1978 to 31 March 1978, (The University of Salford Industrial Centre). Crossley,T. R., Davies, B. J., and Richardson,B. D., 1979,Progress Report, Preliminary Design of a Job Shop Control System,I January 1979 to 31 March 1979, (The University of Salford Industrial Centre) Cullinane,T., and Mayer, R. J., 1992, IDEFO and IDEF3: Methods for implementinga Business warehousecontrol system,Proceedings of process design -A strategyfor corporate information management,October, (IDEF Users Group, Kettering, Ohio). Curwen, P., 1990, Systemdevelopment using the CORE method, BAe-WIT-ML-GEN- SWE-512 and 1227, (British Aerospace(Military Aircraft) Ltd) Cutts, G., 1991, SSADM Structured SystemsAnalysis and Design Methodology, (Black well Scientific, Oxford, UK). 180 Davis, W. J., and Jones,A. T., 1989, A functional approachto designingarchitectures for ML IEEE Transactionson Systems,Man and Cybernetics,19 (2), 164-173. Deasley,P., and Collins, J., 1994,Talk to the City, Manufacturing Engineer, 73 (2), 68-70. DeMarco, T., 1978,StructuredAnalysis and SystemSpecification, (Yourdon Press,NY). Deng, Z., Wang, K., Huang, B., and Bu, E., 1990, A study of different modelling methods for CIMS and FMS, Optimisation of manufacturing systemsdesign, Ed D. L. Shunk, 133-148,(Elsevier, North-Hofland). DiCesare,F., Harhalakis,G., Proth, J., Silva, M., and Vernadat,F. B., 1993,Practice of Petri Nets in manufacturing, (Chapmanand Hall). Doumeingts,G., 1984,Methode GRAI: methodede conceptiondes systemsen productique. Thesed'etat, Universite de Bordeaux, Novernbre. Doumeingts,G., 1985,How to decentralisedecisions through GRAI model in production management,Computers in Inclustry, 6,501-504. Dourneingts,G., and Breuil, D., 1987, The GRAI approach,Proceedings of the A UGRAI meeting of September1987, (The GRAI Laboratory, The University of Bordeaux). Drucker, P. F., 1991, The new productivity challenge,Harvard BusinessReview, November-December,1991,69-79. Mott, L., and Beavis, S., 1994, Yhe Guardian, 7th November. Elliott, L., 1995, Made in Britain, Yhe Guardian, 30th January. Emery, F. E, and Trist, E. L, 1959, Socio-TechnicalSystems, Presented at the 61h International meeting of the Institute ofManagement Science,September, Paris, (PergamonPress, London). EPSRC, 1994, Innovative Manufacturing Initiative, ResponsiveProcessing, (EPSRC, Polaris House, Swindon). Espejo, R., 1989a,The VSM revisited, Yhe Viable SystemsModel: Interpretations and applications of Stafford Beer's VSMed. R. Espejo & R. Harde, 77-100, (Wi1ey, Chichester). Espejo, R,. 1989b,P. M. Manufacturers:the VSM as a diagnostictool, Yhe Viable Systems ModeL Interpretations and applications of Stafford Beer's VSM ed. R. Espejo&R. Harde, 103-121(Wiley, Chichester). Espejo, R., and Hamden, R-, eds., 1989, Yhe Viable SystemsModel: Interpretations and applications of Stafford Beer's VSM, (Wi1ey,Chichester). 181 EswaraReddy, C., KrishnaiahChetty, 0. V., and Chaudhuri,D., 1992, A petri net based approachfor analysingtool managementissues in FMS, International Journal of Production Research,30 (6), 1427-1446. Evershiem.W., and Shultz, J., 1985, Survey of CAPP systems,Annals of CIRP, 35 (2), 607- 613. Feller, A., and Rucker, R., 1990,Extending structured analysismodelling with AL an applicationto NWH profiles and SFC data communicationsrequirements specifications,171-194, Optimisation ofManufacturing SystemsDesign, ed. D. L. Shunk, (Elsevier SciencePublishers B. V., North Holland). Firesmitk D. G., 1993, Object oriented requirementsanalysis and logical design,(John Wiley, NY). Flood, R. L., and Carson,E. R., 1988,Dealing with complexity,An introduction to the theory and application ofsystemsscience, 2 1, (PleniumPress, New York). Frank, M., 1993, CASE for chent/server,DBMS, 6 (3), 34-39. Gane,C., and Sarson,T., 1977, Structured systemsanalysis: Tools and techniques,(IST Inc., New York). Gamble,J. D., 1988, The developmentof a computerbased IDEFo system,MSc dissertation, (Liverpool Polytechnic). Genrich, H. J., and Lautenbach,K., 198 1, Systemmodelling with high level Petri Nets, YheoreficalComputer Science, 13,109-13 6. Godwin, A. N., Gleeson,I W., and Gwillimn T, D, 1988, Yheintegration ofIDEF descriptions, IMMS-8801, (Coventry LanchesterPolytechnic, Coventry). Godwin, A. N., Gleeson,J. W., and Gwilliam T, D., 1989, An assessmentof the IDEF notations as descriptivetools, Information Systems,14 (1), 13-28. Goldman,D. S., and Cullinane,T. P., 1987, AdvancedFactories using Functional Models, Proceedingsof 241hAnnual ConferenceAIM Tech.,Framlington M. A, 4349, (Manufacturing Institute, Beloit, WI, USA). Goosen,A., 1994,Re-engineering the South African Air Force logistics operations, Proceedings Conference 'Beyond of the re-engineering, La Jolla, October, 41 -54, (IDEF Users Group, Kettering, Ohio). Gould, P., and Dent, A., 1993,Research on integration in manufacturingcompanies, A Strategy discussionpaper, (The ACME Directorate, August, 1993). Graefe,U., and Thomson,V., 1989, A referencemodel for production control, International Journal of ComputerIntegrated Manufacturing, 2 (2), 86-93. 182 Graham,1., 1991, Object OrientedMethods, (Addison-Wesley). GRAI Laboratory, 1987,Research Activities, (The GRAI Laboratory, The University of Bordeaux). GRAI Laboratory, 1991, GRAIMethod, The rules of the method and analysisof GRAI Grids' section 14, (The GRAI Laboratory, The University of Bordeaux). Graves,G. R., Sahay,A., and Charles,M. P., 1989, A functional approachto feature recognition in production planning, ComputerIntegratedManufacturing Systems,2 (3), 162-171. Gregory, W. W., 1994,Business driven Re-Engineering,Coupling businessplans and IDEF models,Proceedings of the Conference 'Beyondre-engineering, La JoUa,October, 80-94, (IDEF Users Group, Kettering, Ohio). Gutkowski, S. J., 1994,Enterprise Reengineering, September 94,1, (TFG Publishing, Vienna, VA, USA). Hammer,M., 1990,Reengineering work: don't automate,obliterate, HarvardBusiness Review,July -August, 1990,104-112. Harrison, C. S., Gowland, T., Dove, L. G., and Makin, B., 1988, A practical approachto CIM systemdesign, Proceedings of 41hInternational Conferenceon Computer Aided Production Engineering, Edinburgh,November, 91-94, (Edinburgh University). Harrington, J., 1984, Understandingthe Manufacturing Process, (Marcel-Dekker, NY). Harrington, J., 1985,Understanding the Manufacturing Process-Anapplication of ICAMs IDEF Methodology, Proceedingsof CINCOM 85 Conference,1-17, AnahiernCA., USA., April, (Computer and Automated SystemAssociation of SME, Dearborn, MI, USA) Hartrum, T. C., Connally, T. D., and Johnson,S. E., 1988, An interactive graphicseditor with integrateddata dictionary for IDEFo structured analysisdiagrams, Proceedings of the IEEE 1988 National AerospaceandElectronics Conference,766-770, NAECON, Dayton OIL (EEEE,New York, USA). Hill, S. C., 1995,A conciseguide to the IDEFo technique:A practical techniquefor BusinessProcess Reengineering, (Enterprise Technology ConceptsInc., Puyallup, WA, USA). 11itomi,K., 1990, Manufacturing systemsengineering: The concept, its context and the state of the art, International Journal of ComputerIntegrated Manufacturing, 3 (5), 275-288. Fliton-ii,K., 1994,Manufacturing systems:past, presentand for the future, International Journal ofManufacturing SystemDesign, 1 (1), 1-17. 183 HMSO, 1995,Britain 1995, an official handbook, 230-23 1, OMSO). Ho, Kah-Fei., and Ridgeway,K., 1994, Design of cellular manufacturingsystems using GRAI methodology,Advances in Manufacturing Technology VHI, ed. K. Caseand S. Newman, 169-173,(Taylor and Francis). Holmberg, B, A, 1989,Developing organisationalcompetence in a business,The Viable SystemsModeL Interpretations and applications of Stafford Beer's VSM ed. R. Espejo & R. Harde, (John Wiley & Son, Chichester). Hori, S., 1972,Human-directed activity cell model, CAM-I long range planningfinal report, (CAM-1 Inc.). Hubei, H., and Colquhoun, G. J., 1996, Engineeringdata control for bid and order handling in the capital plant industry, International Journal of ComputerIntegrated Manufacturing, Acceptedby referees- awaiting publication date. Huff, B. L, Liles, D. H, Howell, B. J., Sanders,F. K and Gaddis,B. P., 1991a, The use of EDEFmodelling techniquesfor small manufacturingenterprise modelling, Proceedingsof the Conference'Towards an International framework', Albuquerque,April, 155-169,(IDEF Users Group, Kettering, Ohio). Huff, B. L., Howell, B. J., Liles, D. H., 1991 b, The use of IDEF3 and Coloured Petri Net modelling to support continuousenterprise process improvement, Proceedings of the AdvancedModellingMethods TechnologySymposium, October, 147-163, (Automation and Robotics ResearchInstitute, Fort Worth, Texas.). Hughes,D. R., and MauU,R_, 1985a,Software-the key to computer integrated manufacturing,Proceedings of the Effective CAD CAM Conference,C 137/8 5,5 -22, (IMechE.). HughesD. R., and Maull, R., 1985b,A framework for designof CIM systemarchitecture, Computersin mechanical engineering, September,34-37. Hughes,D. R., and Baines,R. W., 1985,Evolutionary designof computer-integrated manufacturingsystems, Proceedings of the Ist National Conferenceon Production Research,(University of Nottingham). 1BM, 1987, ComputerIntegratedManufacturing. - the IBM experience,18-33 (Findlay PublicationsHorton, Kent, (for IBM UK Scientific and Industrial centre)). IIDEF User Group, 1992, YheIDEFFramework, IDEF-UG-0001 Version 1.5, (IDEF Users Group, Kettering, Mo). International Organisationfor Standards(ISO), 1986, Yhe Ottawa report on reference modelsfor manufacturing standards, Version 1.1, ISO TC I 84/SC5/WGI. (International Organisationfor Standards). 184 Jenks,A., 1993,Wisdom's Wisdom in DoD Reorganisation,Washington Technology, 8 (10), 1, August, (TechNewsInc., Washington). Jennison,L., 1994, Sometools for BusinessRe- Engineering,Software Assistancefor BusinessRe-Engineering, 198-224, (John Wiley & Sons,Chichester). Jensen,K., 1990, Coloured Petri Nets: A 11ighLevel Languagefor SystemDesign and Analysis, Lecture Notes in Computer Science- Advancesin Petri Nets 1990,342- 416, (Springer-Verlag). Jones,C. M., Godwin, A. N., and Gore, M., 1991, Applying the five step method to modelling quality managementsystems, Integration and managementof technology for manufacturing,IEE Management of TechnologySeries 11,250-263, (Peter Peregrinus,London). Jones,J. H., 1988,Making it Happen, 14, (Fontana,London). Jones,R. S., 1979,Management of large scaleintegration during the requirementsphase, Proceedingsof IEEE Computer Society 7hird International ComputerSoftware andApplications Conference,CONTSAC 79, Chicago,IL, November, 337-345. Jorysz,H. R., and VernadatF. B., 1990a,CIM-OSA part 1: Total enterprisemodelling and function view, International Journal of ComputerIntegratedManufacturing, 3 (3 and 4), 144-156. Jorysz,H. R-, and Vernadat F. B., 1990b, CIM-OSA part 2: information view, International Journal of ComputerIntegratedManufacturing, 3(3 and 4), 157-167. KB SL 1994,ProSim Usersmanual and referenceguide version Z 1, (Knowledge Based SystemsIncorporated, College Station, Texas). Keen, A., and Schafiik, F., 1995,IDEN Object-Orienteddesign method (Revision 2 overview), Proceedingsof Charting the Coursefor Enterprise Engineering, June 95, (Society for EnterpriseEngineering, Kettering, Ohio). Kehoe, D. F., Little, D., and Lyons, A. C., 1993, A methodologyto determinethe requirementsfor improved manufacturingintegration, Proceedings ofMethods in Manufacturing SystemsEngineering Conference,May 1993, (The University of Liverpool). Kehoe, D. F., Little, D., Al-Maliki, I, and Wyatt, T. M., 1991, Method for the specification of tool managementinformation systems,Computer-IntegratedManufacturing Systems,4 (1), 18-25. Kief, H. B., and Walters, T. F., 1992, ComputerNumerical Control, (Glencoe,Westerville, Ohio). Kfir, G. J., 1985, Complexity: somegeneral observations, Systems Research, 2, (2), 131-140. 185 Mittich, M., 1990, CIM-OSA part 3: CIM-OSA integrating infrastructure- the operational basisfor integratedmanufacturing systems, InternationalJoumal of Computer IntegratedManufacturing, 3 (3 and 4), 168-180. Klittich, M., 1995, Possible short term benefits from CIM-OSA for users and providers of BPM systems, Proceedings of the 2nd open workshop on Business Process Management based on CIM-OSA, (European Enterprise Integration Supportive Services (EEISS), Eschborn). Komanduri, R-, McGee, J., Thompson,R., Covy, J. P., Truncale, F. J., Tipris, V. A., Stack, K M., and King, K 1., 1985, On a methodology for establishingthe machinetool systemrequirements for high-speed/high-throughputmachining, Transactions of the A&VE Journal of Engineeringfor Industry, 107,November, 316-324. Kusiak, A., Larson, N. T., Wang, J., 1994, Reengineeringof designand manufacturing processes,Computers in Industrial Engineering, 26 (3), 521-536. (Pergamon). Law, M., 1992, Template-drivensystems development with IIDEF:Enterprise standardsfor re-use, Proceedings of 'Businessprocess design -A strateSyfor corporate information management, October 1992,Washington (IDEF Users Group, Kettering, Ohio). Law, A- M., and Kelton W. D., 1991, Simulation modelling and analysis, 234, (McGraw Hill). Le Clair, S. R-, 1982, IDEF the method, Architecture the Means to Improved Manufacturing Productivity, &VE TechnicalPaper, (Society of Manufacturing Engineers, Dearborn, Michigan, USA). Levy, J,. 1993, A recipe for success,Manufacturing Engineer, 72 (6), 258-259. Little, D., 1995,Report of EPSRC workshop on structuredmethods and tools in information systemsspecification and design,6-7, (The Departmentof Industrial Studies,The University of Liverpool) Longworth, G., and Nicholls, D., 1987, Ae SSADMManual, (National Computing Centre, Manchester,UK). Los, PL, Beziau, D., Brodda, J., and Marzi, J., 1992,Development and use of an integrated referencemodel for one-of-a-kind construction site manufacturing, 'One -of-a-kind' production: New approaches,B. E. HischandKD. Yhoben(Eds. ), 329-340, (Elsevier,North Holland). Macconaffl,P., 1990,Introduction to the ESPRIT project, Ae International Journal of ComputerIntegratedManufacturing, 3 (3 & 4), 140-143. 186 MacIntosh, R., 1993,Material flow issuesin manufacturingsystems integration, Proceedings of Methods in Manufacturing SystemsEngineering Conference,May, (The University of Liverpool). Mackulak, G. T., 1984,High level planning and control: An IDEFo analysisfor airframe manufacture,Journal ofManufacturing systems,3 (2), 121-133. Maji, R. K., and Stevenson1. A., 1988, Design Methodologies for CM Proceedingsof IS,4 TA 191hInternational Symposiumon, 4utomotive Technology,Monte Carlo, Monaco, October, 3-19, (Automotive Automation, Croydon, UK). Malhotra, R., and JayaramanS., 1992, An integration framework for enterprisemodelling, Joumal ofManufacturing Systems,11 (6), 426-441. Mandel, K., 1990, Graphicalprocess descriptions-views and diagrarns,International Journal of ComputerIntegratedManufacturing, 3 (5), 314-327. Marca, D. A., and McGowan, C. L., 1987, SADT Structured Analysis and Design Technique,(McGraw Hill, NY). Marran, L., Fadali, M. S., and Tacker, E. C., 1989, A new modelling methodologyfor large scalesystems, Proceedings of the international conferenceon SystemsMan and Cybernetics,5 (2), 989-990, (IEEE, Piscataway,NY). Martin, J., and McClure, C., 1985,Diagramming techniquesfor analystsandprogrammers, 131-135,(Prentice-Hall, Englewood Cliffs, NJ). Maull, R. S., 1986,An evaluation of the contribution of the ICAMDefinition Method- IDEFo to the analysis and design of computer integrated manufacturing systems, PhD thesis,Bristol Polytechnic. Maull, R., Childe, S., Bennett, J., Weaver, A., and Smart, A., 1995,Report on goodpractice in BusinessProcess Re-engineering in manufacturing Wust?y, Working paper WP/GR/J95010-2,(University of Plymouth). Mayer, R. J., 1979,Unified SDM: The ICAM approachto systemsand software development Proceedingsof IEEE ComputerSociety AirdInternational ComputerSoftware andApplications Conference,CONTSAC 79, Chicago, EL, November, 331-336. Mayer, R. J., 1990a,An overview of knowledgebased systems, automated IDEF tools and IDEF services,Proceedings of the conferenceon Enterprise analysis: modelling, integration and maintenancestrategiesfor standard methodologies,Washington, May, 505 -534, (IDEF Users Group, Kettering, Ohio). Mayer, R. J., ed., 1990b,IDEFIx Data modelling, (Knowledge basedSystems, Incorporated, College Station, Texas). 187 Mayer, & J., Cullinane,T. P., DeWitte, P. S, Knappenberger,W. B., Perakath,B., and Wells, M. S., 1992a,Information Integrationfor Concurrent Engineering (IICE) IDEF3 Processdescription capture method report, AL-YR-1992-0057, (Armstrong Laboratory, Wright PattersonAir Force Base, Ohio). Mayer, R. J., Wells, M. S., and Keen, A. A., 1992b,Information Integrationfor Concurrent Engineering (IICE) IDER Object-Oriented design methodreport, Interim technicalreport, AL-TR-1992-0056, (Knowledge Based SystemsIncorporated, College Station, Texas). Mayer, R. J., Griffith, P. A., Menzel, C. P., and Painter, M. K., 1992c,Design rationale capture,Method conceptpaper AL-TP-1992-0050, (Armstrong Laboratory, Wright PattersonAir Force Base, Ohio). Mayer, R. J., ed., 1993a,IDF-Fo Function Modelling, (Knowledge basedSystems, Incorporated, College Station, Texas). Mayer, R. J., ed., 1993b,IDF-F3 Processdescription capture method, (Knowledge Based SystemsIncorporated, College Station, Texas). Maylor, H., and Butler, C., 1991, Total quality control in flexible manufacturingsystems, Proceedingsof the 71hNCPJý Hayield Polytechnic, September,490495, (Hatfield Polytechnic). Menzel, C. P., Mayer, R., and Painter, M. K., 1992, Ontology descriptioncapture method: Conceptsand formal foundations,Interim technical report AL-TP-1992-0051, (Armstrong Laboratory, Wright PattersonAir Force Base, Ohio). Miller, J. G., 1978,Living Systems,(McGraw Hill, New York). Ming Wang, and Smith, G. W., 1988, Modelling CIM systemsPart 1: Methodologies, ComputerIntegratedManufacturing Systems,1 (1), 13-17. Mitchell, F. H., 1991, CAf systems:An introduction to ComputerIntegrated Manufacturing, (Prentice-Hall,London). Moder, J. J., and Phillips, C. R., 1970,2nd edition, Project managementwith CPM and PERT, (Van Nostrand Reinhold). Molina, A., Ellis, T. I. A., Young, R. I., and Bell, R., 1995,Modelling manufacturing capability to support concurrent engineering,7he Journal of Concurrent engineering.- researchand applications, 3 (1), 2942. Mujtaba, M. S., 1992, Systemswith complex material and information flows, Proceedingsof the international conferenceon Object-Oriented manufacturing systems,May, 188- 199, (University of Calgary,Alberta, Canada). 188 ujtaba,M. S., 1994, Simulat on modelling of a manufacturingenterprise i&ith complex material,information and control flows. International Journal of Computer IntegratedManufacturing, 7 (1), 2946. Nassi, I., and Schneiderman,B., 1973, Flowchart techniquesfor structured programming, ACMSIGPLANNotices, 8 (8), 12-26. Niel, E., Zaytoon, J., Mille, A., and Jutard, A., 1992, A contribution for integrating the operationalsafety concept in CEýLInternational Journal of ComputerIntegrated Manufacturing, 5 (6), 349-360. Nijssen, G. M., and Halpin, T. A., 1989, Conceptualschema and relational database design, afact oriented approach, (PrenticeHall). O'Connor, P. D. T., 1991, Practical Reliability Engineering, 146-162,(Wfley, NY). Oppenheirn,A. N., 1970, Questionnairedesign and attitude measwrement,(Heinemann EducationalBooks, London). O'Sullivan, D., and Browne, J., 1993, Integrated Manufacturing SystemsDesign Procedure, Advancesin ComputerIntegratedManufacturing, Ed. P. J. Sackett,4-35, (JAI Press,London). Page,C. J., and Saunders,M. J., 1993, 'The Five Step Method': A framework for defining and developingcomplex manufacturingmanagement systems, Innovation and Product Development,Ed P.J. Nolan, Vol. 2,753-764, (Galway University Press). Painter, M. K., 1992, Idormation Integration for ConcurrentEngineering (IME), a foreword to Technical report AL-YR-1992-0057,(Armstrong Laboratory, Air force SystemsCommand, Wright PattersonAir Force Base). Pandya,K. V., Sotudeh,R., and Bunn, P. R., 1990, A model of cell supervisordecision support systems,Proceedings of the 7th International conferenceon mathematical and computermodelling, Vol. 14,1040-1045, (Pergamon). Parnaby,I L., 1988, YheLucas Manufacturing SystemsEngineering Handbook, 2.4, (Lucas EngineeringSystems, SolihuU, UK). Perakath,C. B., 1995, In answerto questionsduring presentationof, Toward an environrnentfor coUaborativeenterprise engineering (CEE), Proceedingsof Charting the Coursefor Enterprise Engineering, June, (Society for Enterprise Engineering,Kettering, Ohio). Perakath,C. B., Mayer, K J., Blinn T. M., Marshall, C., and Tye, R., 1995, Toward an environmentfor collaborative enterprise engineering (CEE), Proceedingsof Charting the Coursefor Enterprise Engineering, June, (Society for Enterprise Engineering,Kettering, Ohio). 189 Peterson,I L., 1981, NO Net 7heory and the Modelling ofSystems, (PrenticeHaU, London). Petri, C. A., 1980,Introduction to generalnet theory, Net theory and applications, ed. W. Brauer, (Springer, Berlin). Preston,R., 1990,Executive overview, Enterprise analysis:Modelling, Integration and Maintenancestrategies for standardmethodologies, Proceedings ofIDEF Users Group Conference,May 1990,3-27, OIDEFUser Group, Kettering, Ohio). Preston,R., 1995, Charting the coursefor the future SEE strategic direction, Chairman's addressto the SEE conference'Charting the coursefor EnterpriseEngineering', Orlando, Florida, USA, (Society for EnterpriseEngineering Kettering, Ohio). Pritsker, A. A. B., 1986,Simulation and SLOM, (JohnWfley & Sons,NY). Pritsker, A. A. B., 1974, The GERTE usersmanual, (Pritsker and Associates,Inc., West Lafayette, IN). Pun, L., Doumeingts,G., and Bourley, A., 1985, The GRAI approachto the designof Flexible Manufacturing Systems,International Journal of Production Research,23 (6), 1197-1215. Ranky, P. G., 1992,Flexible Manufacturing Cells and Systemsin CIM, (CIMware Ltd, Gufldford, UK. ). Rembold,U., Nnaji, B. 0., Storr, A., 1993, ComputerIntegratedManufacturing and Engineering, (Addison-Wesley,Wokingham). Ridgeway,K., Downey, S 1991, A techniquefor the introduction and -, graphical planning of AMT: The GRAI method, Integration and managementof technology, eds.E. H. Robson,H. M. Ryan and D. Wilcock, 55-65, (Peter Peregrinus). Ridgeway,K., 1992, Analysis of decisioncentres and information flow in project management,International Journal ofProject Management, 10 (3), 145-152. Roboam,M., and Pun, L., 1989,Utility of designmethodology for advancedmanufacturing systems,Computer IntegratedManufacturing Systems,2 (1), 4-10. Roboam,M., Zanettin, M., and Pun, L., 1989, GRAI-I]DEFO-Merise(G": Integrated methodologyto analyseand designmanufacturing systems, Computer Integrated Manufacturing Systems,2 (2), 82-98. Rochfeld, A., and Tardieu, H., 1983,MERISE: An information systemdesign and developmentmethodology, Information andManagement. 6 (3), 143-149. Ross, D. T., 1977a,Reflections on requirements',Guest editorial, IEEE Transactionson SoftwareEngineering, SE-3, (1), 2-5. 190 Ross,D. T., 1977b, Structured analysis(SA): A languagefor communicatingideas, IEEE Transactionson SoftwareEngineering, SE-3, (1), 16-24. Ross, D. T., 1985, Applications and Extensionsof SADT, Computer, April, 25-34. Ross, D. T., 1994, SADT-SIG presentation,Proceedings of Beyond Re-engineering' Conference,19, (IDEF Users Group, Kettering, Ohio). Ross,D. T., and Schoman,K. E., 1977, Structured analysisfor requirementsdefinition, IEEE Transactionson SoftwareEngineering, SE-3, (1), 6-15. Ross, D. T., and Schoman,K. E., 1979, Structured analysisfor requirementsdefinition, Classicsin SoftwareEngineering, ed. E.N. Yourdon, (Yourdon Press,NY). Ross, D. T, Brackett, J. W., Bravoco, R. R., and Schoman,K. E., 1980,Architects manual ICAM definition melhodIDEFO', Report No. DR-80-ATPC, (CAM-I, Arfington, Texas). Rumbaugh,J., Blaha, M., Premerlani,W., Eddy, F., and Lorensen,W., 1991, Object- Oriented modelling and Design, (Prentice-Hall,Englewood Cliffs, NJ). Rzevski, G., and Maji, K K., 1988, Computer aidedproduction managementfor bare board manufacture,Proceedings of 4th International Conferenceon ComputerAided Production Engineering, 169-172,Edinburgh, November, (Edinburgh University). Sarnuels,R. J., 1994,Pathways of technologicaldiffibsion in Japan,Sloan Management Review, Spring, 1994,21 - 34. Sarkis, J., and Lin, L., 1994, An IDEFo functional planningmodel for the strategic implementationof CIM systems,International Journal of ComputerIntegrated Manufacturing, 7 (2), 100-115. Sauter,I A., and Judd. R. P., 1990, Applying integratedtools to the designof manufacturingsystems, Proceedings ofA UTOFACT 90 Conference,251 - 25_17, M, USA. Scheer,A. W., 1992, Architecture of integratedinformation systems,ARIS, (IDS, Prof. ScheerGmbI-L Saarbrucken). Schneider,J., 1994,Developing prototype information systemsfrom IDEFo, IDEFIx and IDEF3 models,Proceedings of 'BeyondRe-Engineering'Conference, 183-196, (The IDEF Users Group, Kettering Ohio). Shlaer,S., and Mellor, S., 1988, Object-Orientedsystems analysis: Modelling the real world in data, (PrenticeHall, Englewood Cliffs, NJ). Shunk,D., Sullivan,B., and Cahill, J., 1986,Making the most of IDEF modelling-TheTriple Diagonal concept, CIMReview, Fall, 2-17. 191 Siggaard,K., and Bilberg, A., 1992, Simulation for shop floor control, Proceedingsof CIRP International Seminaron Manufacturing Systems,383-394, (TechnicalUniversity of Denmark). Silva, M., 1993,Introducing Petri Nets, 1, Practice ofPetri Nets in manufacturing, (Chapmanand HaH). SmaU,C., Baines,R. W., and Colquhoun, G. J., 1996,Bestpractice modelling methods, preliminary results, (EvIRC,PO Box 195, Derby). Smith, G. L., Stephens,S. A., Tripp, L. L., and Warren L. W., 1980, A tool to support designautomation in batch manufacturing,Proceedings of ACM Design Automation Conference,Minneapolis, June, 367-373. Snyder,D. E., McNeese,M. D., and Zaff, B. S., 1991,Identifying designrequirements using integratedanalysis structures, Proceedings ofIEEE 43rdAerospace and Electronics Conference,NAECON, 779-785, (IEEE, Piscataway,NY). Soff ech Inc., 1980,Architects Manual, ICAMDMnition Method, 'IDEFO', DR-80-AT? C- 01, (Soffech Inc. Waltham, Ma). Sommerville,1., 1992,Software Engineering, 4th Edition, (Addison-Wesley,Wokingham). Son, S. K., Choi, B. K., Park, T. U., and Lee-Kwang, H., 1991, Resolution of multiple conflicts in the Petri net model of FAS, International Journal of Computer IntegratedManufacturing, 4 (4), 253-262. Stephans,D., and Fox, D., 1987,IDEF modelling applicationsto quality assurance assessments,quality the universalequation for excellence,Proceedings of the American Societyfor Quality Control 41st Annual Quality CongressTransactions, 512-517, Milwaukee, WI, USA, (ASQC, Milwaukee, WI, USA). Stockenberg,I E., 1986, A unified methodology for developingsystems, Proceedings of the IEEE ComputerSociety's International SoftwareandApplications Conference, Chicago,274-276. (IEEE). Sudnick, D. R., 1993,IDEFO - To an International Standard, (The IDEF Users Group, Kettering, Ohio). Sutton, G. P., 1989, Survey of processplanning practices and needs,Journal of Manufacturing Systems,8 (1), 69-7 1. Tannock, J. D. T., and Maull, R. S., 1988, The integratedquality systemin CRvLComputer IntegratedManufacturing Systems,1(4), 228-234. Tran, M., 1993, AmericanNotebook, 7he Guardian, 8/11/93. Triune Software, 1993,Restructuring processes for profitability, (Triune Software, Fairborn, Ohio). 192 US Air Force, 1974,Air Force ComputerAidedManufacturing (AFCAM) Master Plan, Air Force Materials Laboratory, Wright-PattersonAFB, Ohio 45433, AFNIL-TR-74- 104, July. USAir Force; 1981a, Integrated Computer-AidedManufacturing (ICAM) Architecture Part 11,Volume IV- Functional Modelling Manual (IDEFO), Air Force Materials Laboratory, Wright-PattersonAFB, Ohio 45433, AFWAL-IR-814023, June. US Air Force, 1981 b, Integrated Computer-AidedManufacturing aC4M) Architecture Part 11,Volume V- Information Modelling Manual aDEFI), Air Force Materials Laboratory, Wright-PattersonAFB, Ohio 45433, AFWAL-TR-81-4023, June. US Air Force, 1981 c, Integrated Computer-AidedManufacturing (ICAM) Architecture Part 11,Volume P7-Dynamics Modelling Manual (YDEF2),Air Force Materials Laboratory, Wright -PattersonAFB, Ohio 45433, AFWAL-IR-81-4023, June. US Air Force, 1981 d, Integrated Computer-AidedManufacturing (ICAM) Architecture Part II, VolumeIV, Architecture a structured approach to manufacturing, Air Force Materials Laboratory, Wright-PattersonAFB, Ohio 45433, AFWAL-TR-814023, May. US Air Force, 1981 e, Integrated Computer-AidedManufacturing aCAM) Architecture Part II, VolumeM, ICAM library maintenanceand distribution procedures, Air Force Materials Laboratory, Wright-PattersonAFB, Ohio 45433, AFWAL-TR-81-4023, May. US Air Force, 198If, Integrated Computer-AidedManufacturing (7CAM)Architecture Part H, VolumeX, dynamicsmodel of a sheetmetal centre sub-system,Air Force Materials Laboratory, Wright-PattersonAFB, Ohio 45433, AFWAL-TR-81-4023, May. US Air Force, 1986,Integrated Computer-AidedManufacturing (ICAM) Architecture Part IV, Volume V-Information Modelling Manual (7DEF] Extended),Air Force Materials Laboratory, Wright-PattersonAFB, Ohio 45433, AFWAL-TR-864006, June. Wade, M., and Karaszewski,Z. J., 1992,Infrastructure study in sMpbuilding:A systems analysisof US commercialsMpbuilding practices, Journal of Ship Production, May 1992,111 - 126. WaIdner,J., B., 1992,Principles of ComputerIntegratedManufacturing, 120-124(WHey, New York). Wang, Ching-Yang, and Fulton, R. E., 1994, Information systemdesign for optical fibre manufactureusing an object-orientedapproach international Journal of Computer IntegratedManufacturing, 7 (1), 61-73. 193 Wang, H. P., and Wysk, R. A., 1988, AIMSI; a prelude to a new generationof CAD-CAM systems,International Journal ofProduction Research,26 (1), 119-131. Wang, W., Popplewell,K., and Bell, R., 1993, An integratedmulti-view systemdescription approachto approximatefactory modelling, International Journal of Computer IntegratedManufacturing, 6 (3), 165-174. Ward, P. T., 1986, The Transformation Scheme:An extensionof the data flow diagramto representcontrol and timing, IEEE Transactionson SoftwareEngineering, SE-12, 2,198-210. Wamier, J., 1981, Logical construction ofsystems, (Van Nostrand Reinhold, NY). Weston, R- H., 1995,An academicperspective on structured methodsand toolsfor the life- cycle engineering of information systems,(MSI ResearchInstitute, Loughborough University of Technology). Wflson, B., 1984,Systems: Concepts methodologies and applications, Wfley, (New York). Wisdom Systems,1995, WashingtonTechnoloSy, (Wisdom Systemsnews-sheet). Wisnosky, D. E., 1979, ICAM - Foundation for Next GenerationFactories, Industrial Engineering, April, 3845. Wood, P. J., and Johnson,P. N., 1989, A review of the use of SSADM and IIDEF at the University of Warwick, Proceedingsof SAMT 89 (SunderlandPolytechnic, UK). Wu, B., 1992,Manufacturing SystemsDesign andAnalysis, (Chapmanand HaU,London). Wu, B., 1994,Manufacturing SystemsDesign and Analysis, 2nd ed Context and Techniques,(Chapman and HaU,London). Wyatt, T., and AI-Maliki, 1., 1990,Methods in manufacturingsystems engineering-the background,IntegratedManufacturing Systems,1 (2), 91-93. Yadav, S. B., 1983,Determining an organisatiotfsinformation requirements:a state of the art survey,Database, Spring, 3-20. Yadav, S. B., Bravoco, R. R., ChatfieldýA. T., and Rajkusar,T. M., 1988, Comparisonof analysistechniques for information requirementsdetermination, Communications of A CM., 31 (9), 1090-1097. Yancey, D. P., 1985, An overview of spacesystem modelling using high-level process simulationlanguages, SAýE lechnicalpaper, MS85-983, (Society of Manufacturing Engineers,Dearborn, Nfl). Yeomans,P. H., 1987a, 7he benefitsof IDEFo in modelling manufacturing operations, (NEcroMatchLtd, Crowthome, UK). 194 Zachman, I A., 1987, A framework for information systems arcMecture, IBM Systems Journal, 26 (3), 276 -292. Zaff, B. S., and McNeese,M. D, 1991, An integratedmethodology for knowledge and designacquisition, Proceedings of IEEE 43rd Aerospaceand Electronics ConferenceNAECON, 779-785, (IEEE, Piscataway,NY). Zgorzelski, M., and Zgorzelska,P., 1993, IDEF and NLAM: on the convergenceof two outstandingmethodologies, Proceedings of IDEF and the Corporate Infrastructure conference,University of Maryland, 87-97, (IDEF Users Group, Kettering, Ohio). Zhang, H., and Alting, L., 1994, Computerisedmanufacturingprocessplanning systems, (Chapmanand HaU,London). Zhang, W., 1989.Representation of assemblyand automatic robot planningby Petri net, IEEE Transactionson Systems,Man and Cybernetics,19 (2), March/April, 418- 422. Zhao, Z., Baines,R. W., and Colquhoun, G. J., 1995, Infonnation integration of CAD-CAM using IDER, Advancesin Manufacturing TechnologyK YheEleventh National Conferenceon Manufacturing Research,De Montfort University (Taylor and Francis). 195 BIBLIOGRAPHY Adiga, S., 1993, Object - Oriented Software for Manufacturing Systems, (Chapmanand HaU,London). Askin, R. G., and Standridge,C. R., 1993, Modelling and analysisof manufacturingsystems, (Wiley, New York). Barwise, J., 1983, Situationsand Attitudes, (The MIT Press,Cambridge, Massachusetts). Bertalanfly, L. G., 1971, Generalsystems theory, foundations,development, applications,(Allen Lane, The PenguinPress, London). Cooper, R., Mukai, K., and Perry, P., eds., 1990, Situation theory and its applications,(Centre for the Study of Languageand Information, SRI International,Menlo Park, CA). Devlin, K., 1991,Logic and Information, (CambridgeUniversity Press, Cambridge). Laslo, E., 1972, Introduction to SystemPhilosophy - toward a new paradigmof contemporarythought, (Gordon and Breach SciencePublishers, New York). Law, A. M., and Kelton, D. W., 1991, Simulation modelling and analysis, 2nd ed. (McGraw-Hill). Nicholson, A., Coughlan,P., and Syms,H., 1992, Customerservice, productivity and processdesign: impact and attribution, International Operations, Crossing borders in manufacturing and service, (Elsevier Science PublishersB. V. ). Papadopoulos,H., Heavey, C., and Browne, J., 1993, Queueingtheory in manufacturingsystems analysis and design,(Chapman and Hall, London). Reisig, W., 1982, Petri Nets, an introduction, (Springer-Verlag,Berlin). Rosenberg,G., ed., 1991, Lecture notes in computer science,Advances in Petri Nets 1990, (Springer-Verlag,Berlin). Rosenhead,J., ed., 1989, Rational analysisfor a problematicworld, problem structuring methodsfor complexity, uncertainty and conflict, (John Wiley & Sons,Chichester). 196 Senge,P. M., 1990, The Fifth Discipline, (Doubleday,New York, NY). Showa,J. F., 1984, ConceptualStructures, (Addison-Wesley, Reading, Ma. ). Shneiderman,B., (ed.), 1978,Databases: Improving usability and responsiveness, (AcademicPress, New York). Spurr, K., LayzeU,P., Jennison,L., and Richards,N., eds., 1993, Software Assistancefor BusinessRe-Engineering, (John Wiley & Sons, Chichester). 197 APPENDIX I PAPERS PUBLISHED IN THE COURSE OF THE RESEARCH 198 Publications directly related to the research (abstracts are presented in appendix 12): Colquhoun,G. J., Gamble,J. D., and Baines,R. W., 1989, The Use of IDEFOto Link Design and Manufacture in a CIM Environment, International Journal of Operations and Production Management,9 (4), 48-65. Baines, R. W., and Colquhoun, G. J., 1989, The use of a generic EDEFomodel to capture process planning functional informatiod, Proceedings of Sunderland Advanced Manufacturing Technology International Conference, Sunderland, March, 8A/4, (SunderlandPolytechnic) also published in Baines, R. W., and Colquhoun, G. J., 1991, The use of a generic 11DEFomodel to capture process planning functional information, Integration and Management of Technologyfor Manufacturing, IF-E Management of Technology Series 11,3 71- 381, (Peter Peregrinus,London). Baines, R. W., and Colquhoun, G. J., 1990, II)EFo an Integration and Design Tool for Engineers,Assembly Automation, 10 (3), 141-145. Colquhoun, G. J., and Baines, R. W., 1991, A Generic IIDEFOModel of ProcessPlanning, International Journal ofProduction Research,29 (11), 2239-2257. Baines, R. W., Zhao, Z., and Colquhoun, G. J., 1992, A comparisonof two approachesto generative computer-aided process planning (CAPP) using a graphical structured analysis technique, Proceedings of Sunderland Advanced Manufacturing TechnologyInternational Conference,Sunderland, April, (SunderlandPolytechnic). Colquhoun, G. J., and Baines,R. W., 1992, DDEFo,A review of practical considerationsin its use, Proceedings of 81hInternational Conferenceon Computer-Aided Production Engineering, Edinburgh, August, 132 - 136, (Edinburgh University). Colquhoun, G. J., Baines,R. W., and Crossley,R., 1993, A state of the art review of IDEFo, Intenzational Jounial of ComputerIntegratedManufacturing, 6 (4), 252 - 264. Colquhoun, G. J., Baines, R. W., 1993, A comparative investigation of manufacturing systems analysis techniques, Proceedings of the Tenth Conference of the Irish Manufacturing Committee, 492-502, University College Galway, (Officina Typografica, Galway University Press). Colquhoun, G. J., and Baines,R. W., 1993, IDEFo -A survey of use and researchin the UK and Europe, Proceedings of the 1993 IDF-F User Group Conference - Global Competitiveness,Integrating Process, Information and Technology, October, Salt Lake City, Utah, 13 - 28, (IDEF User Group, Kettering, Ohio). 199 Colquhoun, G. J., and Baines, R. W., 1994, A Trocess description approach' to manufacturing systems modelling, Advances in Manufacturing Technology VIH, Proceedings of the Tenth Conference on Manufacturing Research, 360-365 LoughboroughUniversity of Technology, (Taylor and Francis). Hubel, H., Colquhoun, G. I and Hess, P., 1994, An integrated DatabaseSystem for a tender to design to manufacturing system. Advances in Manufacturing Technology VIII, Yhe Tenth National Conference on Manufacturing Research, 5 16-521 Loughborough University of Technology, (Taylor and Francis). Zhao, Z., Baines,R_ W., and Colquhoun, G. J., 1995, Infonnation integration of CAD-CAM using IDEF4, Advances in Manufacturing Technology K Yhe Eleventh National Conferenceon Manufacturing Research, 139-144 De Montfort University, (Taylor and Francis). Hubel, H., and Colquhoun, G. J., 1996, Engineering data control for bid and order handling in the capital plant industry, International Journal of Computer Integrated Manufacturing, Acceptedby referees- awaiting publication date. Hubel, H., Hess, P., Colquhoun, G. J., Hanneghan,M., and Merabti, M., 1996, Concurrent Engineering effizient nutzen, Verein Deutscher Ingenieure (VDI) Zeitschrift, 138 (3), 51-56. Colquhoun, G. J., Baines,R. W., and Crossley,R-, 1996, A compositebehavioural approach to manufacturingsystems modelling, International Journal of Computer Integrated Manufacturing, Acceptedby referees- awaiting publication date. Baines, R. W, SmaU,C., Colquhoun, G. J., 1996, Modelling methods in the manufacturing changeprocess, Yhe 121hNational Manufacturing Research Conference,(NMCR) September1996, The University of Bath. Awaiting refereesreports. Other related publications: Morley, K., and Colquhoun, G. J., 1993, The computer aided design of gear shavingcutters, Proceedings of Yhe International Conference on Engineering Software, StaffordshireUniversity, September,1993, (StaffordshireUniversity). Hanneghan,M., Colquhoun, G. J., and Merabti, M., 1995, Concurrent Engineering support environments, poster paper, 1Ith International Conference on Computer-Aided Production Engineering, June 1995, (The Institution of Mechanical Engineers, London UK). Hanneghan, M., Colquhoun, G. J., and Merabti, M., 1995, Computers in Concurrent Engineering - an enabling technology, poster paper, CoUoquium on 'Concurrent Engineering - Getting it right first time', September 1995, (The Institution of Electrical EngineersLondon, UK). 200 Hanneghan,M., Merabti, M., and Colquhoun, G. J., 1995, The designof an Object-Oriented repository to support concurrent engineering, Proceedings of the Second International Conference on Object-Oriented Information systems. (OOIS'95)., eds.,I Murphy and B. Stone, 200-219, (Springer-Verlag,New York). Hanneghan,M., Merabti, M., and Colquhoun, G. J., 1996, The Worid Wide Web as a platform for supporting interactive concurrent engineering, Lecture Notes in ComputerScience, Paper no. 131, (Springer-Verlag,New York). 201 APPENDIX2 THE IIDEFO-3 METHODOLOGY USER GUIDE 202 CONTENTS A2.1 INTRODUCTION 203 ...... A2.2 IDEFO-3 MODELLING CONCEPTS 204 ...... A2.3 BUILDING IDEFO-3MODELS 207 ...... A2.3.1 ICOM DECOMPOSITION IN AN IDEFO-3 MODEL 207 ...... A2.4 CONSTRUCTING IDEFO-3 DIAGRAMS 216 ...... A2.5 USING TIHE IDEFO-3METHODOLOGY 218 ...... A2.1 INTRODUCTION In developing IDEFO-3 to contribute to enterprise improvement, to gain acceptanceas a routine managementtool and to promote the use of modelling in manufacturing, certain featureswere considered desirable: * It must describe an aspect or aspectsof an enterprise in a way suitable for managing change in manufacturing where a knowledge of functional structure and process behaviourare desirableprior to a changeor enhancement. v It must be capable of capturing and representing the perceptions of participants in a systemin a way that makesit possible to identify qualitative or quantitative differences betweenthose perceptions. This ability is necessaryif a method is to have sufficient resolution to allow meaningful analysis and to provide the potential for optimisation. 7he method should also be capable of aggregation and decomposition but should not encouragethe distortion of a systemdescription tofit the characteristics of the method v Yhe model and its diagrams should be simple enough to be used as a basis of communication betweensystem participants and the analyst. 7he method needs, in its 203 initial form, to be graphical, capable of being produced with or without a CASE tool and use a minimal number of diagrams to complete a model. For intuitive use simple graphical constructs are required negating the needfor an extensiveknowledge of the modelling technique. v Yhe method should require minimal time to gather the information necessary to construct diagrams, to build models and to validate and analyse them. It should be useful in the change process in allowing decision making to take place using more complete and realistic information. Yhe model mustfacilitate analysis and evaluation to assist in predicting the perfonnance and behaviour after a change has beenimplemented The EDEFO-3approach provides these features by using the strengths of the IIDEFo and IDEF3 methods and by correlating IDEFo and IDEF3 diagrams. In the IDEFO-3 methodology no changesare made to the establishedtechniques or syntax used by IDEFo and IDEF3 diagrams (those documentedin US Air Force, 1981a, and Mayer, et al, 1992a respectively). This user guide is concernedonly with defining the conceptsand rules for the correlation of IDEFOand IDEF3. A2.2 IDEFO-3 MODELLING CONCEPTS The IDEFO-3 method can be used to model systemsin manufacturing enterprisesthat can include combinations of software, hardware, machines,facilities, objects and people. For existing systemsIDEFO-3 can be used to investigateand understandfunctional structure and aspectsof behaviour. For new systemdesign it can be used as a common understandingor blueprint for organisationalor systemimplementation. By applying the method a model consisting of a set of diagrams is produced. The model comprises IDEFo diagrams and IDEF3 Process Flow Networks (PFNs) cross-referenced (correlated) using IDEFO-3 diagrams. If necessarya model can also include text descriptions 204 such as the 'page pair' for IDEFo diagrams and 'Elaboration Documents' for IDEF3 diagrams.In additiona nodeindex or nodetree canbe usedto describemodel structure. An IDEFO-3 model separates a system into its constituent parts using structured decomposition to ensure relationshipsbetween all the parts are maintained and traceable. The initial diagrams in a model are the most general, aggregated (i. e. most abstract) descriptionof the system. 11DEFodiagrams are used for the higher level diagramsto describe functional detail and the relationshipsof the information or objects that functions use, need or produce. The analyst applying the method must decide at what stage in the IDEFo decompositionprocess it is necessaryto understandbehavioural aspects of the system. It is at this stagethat the correlation of 11DEFOand IDEF3 PFNs using DDEFO-3diagrams takes place. At this stage,functions (boxes) in IDEFo diagramscan become'scenarios' for PFNs. In IDEFoa functionis characterisedby an activeverb phrase,in generalterms - 'do something' as shownin figure A2.1. The verb can be specific,such as, measure,assemble, select, or conceptualsuch as, evaluate, develop, resolve. DO SOMETHNG DO 'B' DO 'C' Figure A2.1 IDEFo decomposition 205 Figure A2.1 shows conventionalIDEFo functional decompositionwhere the functions B, C and A are shown as the constituent functions of 'do something. No sequenceis inferred by their relative position, the diagram simply describes the situation that 'do something' involves functions B, C and A. If it is necessaryto understandnot just the functional relationshipsbut also the behaviour of 'do something', decompositionof the function to a PFN is necessaryto describe how 'do something' is carried out, figure A2.2 shows the decomposition. In this case A and C are carried out concurrently and start at the same time. They must be completed, but not necessarilyat the sametime, before B can be started. In this decompositionthe aggregated function 'do something' hasbeen described as a set of sub-processesrelated by a sequence. DO SONIETHNG Figure A2.2 H)EFo to PFN decomposition 206 A2.3 BUILDING 11DEF0-3MODELS An IDEFO-3 model is a set of diagrams together with any necessary supporting documentation. A model begins with a viewpoint and purpose declaration and a conventional IIDEFocontext diagram. The single box context diagram is then decomposed successivelyinto more detailed IDEFo diagrams until the stage where it is necessaryto understandor describebehavioural aspects of the systemunder review. This stage could be after one IDEFo decomposition level or after many levels. Experience with the method to date has shown that this stage is characterisedby: (i) a domain experts description moves away from aggregateddescriptions to become a sequentialnarrative (i. e. first we do this, then we wait for A, then X arrives), (H) the use of IDEFo to describe the systemis becoming difficult becauseof the need to force aggregationon a real situation in order to describe it with IIDEFo syntax or (iii) simply that a detailed understanding of behaviouris required. At this stageIIDEFO-3 syntax and rules are used to correlate IIDEFoand IIDEF3 diagramming conventions. A2.3.1 ICOM DECOMPOSITION IN AN IDEFO-3 MODEL IDEFo functions (boxes) represent related sub-functions. The sub-functions can perform various aspects of the larger aggregated function under different circumstances,using different combinations of inputs and controls and producing different outputs. Using correlation rules and syntax IIDEFO-3describes the different activations of the higher level box using an IDEFO-3 diagram. 207 e INPUTS and OUTPUTS INPUT arrows shown entering the side of an IDEFo box are symbols representing data, information, objects or anything that can be described by a noun phrase. INPUTS are necessaryto carry out the function and when the function is performed the INPUT is converted (or processed)into an OUTPUT shown by an outgoing arrow on the right hand side of a box. Figure A2.3 shows a parent IDEFo box with a detailed PFN which can be consideredto fit insidethe parent box. A box representingthe function 'Do something' INPUT OUTPUT --ON. Figure A2.3 IIDEFOparent and PFN chUd In figure A2.3 C and A are processesor UOBs that take part in the transformation of INPUT to OUTPUT they are therefore necessaryfor the processesto be initiated. To describethis the input is shown on an BDEFO-3diagram as a labeHedarrow entering the junction horizontally. The arrow enters or leaves through a 'gate' on the border of an IDEFO-3 diagramto indicate a correlation with a higher level IDEFOdiagram. An exampleof an IDEFO-3 diagram is used to illustrate this feature in figure A2.4. In an IDEF3 model a 208 PFN starts with a singlejunction or process and no separatePFNs are describedin a single diagrain. Rough cut Pmflfing profiling I 00 TO schedules Information (12 wk Plan profile horizon) Procedures releting to: nesb (data Variation to scIlhadulas. L Vanser) nesting priorities planning horizzons.authority Profiling nests Component quantmist due dates (NestIng consVWnts) ýR -an ;a I Varlatonleat: ;o to ek Plan p ch warlatFUlons *sft JI Plate focluiremenis Component porneiry. Release (gauge,no. of pert no. ple plates) (profile, mal.. Ikkness) requirement data Nesbng Informabon (dataGet -Re - kase masonrepartB) p ling clartam viork ctre! RADAN rwMlng Engineer An input 'gate' used to indicate the correlation of the higher level D:)EFo input 'Component geometry, part no.' with the process'Plan profile nests' Figure A2.4 An IIDEFO-3diagram Similarly OUTPUTS in a parent IDEFo diagram are produced by a process or sequenceof processes,so any sequenceof processescan be correlated with an output in the parent IDEFo diagram. This is shown in figure A2.3 where the output from the 'Do something' box is correlatedwith the sequenceof processesculminating in processB. Figure A2.4 shows an IDEFO-3 diagram and the use of output 'gates' to indicate the correlation. In this casethere are three terminating processes,each one has an output which has been describedby the 209 parent IDEFo diagram. Terminating processesare the last processin sequences,figure A2.5 shows three types of terminating processes. The absenceof an output from a terminating process would flag either a correlation problem (such as; is there an output in the parent IDEFo diagram? is the processdescription valid? ) or a potential anomalyin the systemunder investigation. ainatingprocesses (i) terminatingprocess terminating processes (ifi) Figure A2.5 Types of temiinating processes 210 & CONTROLS In an IDEFo diagram a control is distinguishedfrom an INPUT by the assumptionthat an arrow is a control unlessit obviously servesonly as an input, any function must have at least one control. To correlate control arrows between a parent IDEFo diagram (or scenario)and IDEFO-3 diagrams the controls are examined in terms of their influence on junctions (decisions)and processes. In an MEF3 diagramjunctions describethe logic of processbranching which can be either, a process that splits into two or more process paths (fan out junctions), or two or more processpaths converging into a single path (fan in junctions). Junctions can be AND (&), OR (0) or Exclusive OR (X) and synchronous or asynchronousprocess actuation and completion is representedby the junction symbol. For examplein figure A2.5 (i) the fan out junction is an asynchronousAND junction describingthe situation where the two following Processesdo not start simultaneously. In figure A2.5 (iii) the asynchronousOR fan out junction describesthe situation where one or both of the processeswill start, and the synchronousOR fan injunction indicatesthat the processeswill complete simultaneously. To correlate IDEFO and IDEF3 PFNs in an IIDEFO-3diagram it is necessaryto connect EDEFocontrols with processesor junctions in the IIDEF3PFN. The syntax used is shown in figure A2.6. The IDEFO-3 diagram controls (dotted lines) are shown entering through &gates'in the border that connect with either a fan out junction or a link. The diagram describesa situation where: * When the processW is completed,the information,objects or data summarisedby control 'A' must be availablefor the processW to proceed. and e beforeeither process 'Y' or 'Z' (or both processes)can proceed the information,objects or datain the control 'B' mustbe availableand will influencethe logic describedby the ORjunction,i. e. wiH influencewhether either or both processesoccur. 211 gates Control. 'B' Figure A2.6 IIDEFO-3syntax. Controlsare correlatedwith links usingdotted lines and a connectorand they can constrain the following types of 11DEF3links; A 'precedencelink' representingtemporal precedence(the most common link) is shown constrainedby a control. This describesthe situation that the next processis only enabledwhen the information, objects or data describedby the controlare available. 212 A 'relationallink' (or user definedfink) carriesno pre-definedsemantics and is i Do-shown constrained by a control. This describesthe situationwhere a control influencesthe userdefined link. An 'object flow link' describesthe sametemporal semanticsas a precedencelink and also higWights a significant object flow between two Processes. As in the caseof a precedenceUA the use of a control describesthe situation that the next processis only enabledwhen the information,objects or data describedby the controlare available. Controls can influencethe following junction types; Fan out junctions AsynchronousAND - All the following processesstart, but not start simultaneously,The conditions for start are influencedby the control. AsynchronousOR - One or more of the following processeswill start, but will not start simultaneously. The decision or constraint effecting which processesstart is influencedby the control. SynchronousOR - One or more of the following processeswill start simultaneously. The decision or constraint effecting which processes start is influencedby the control. 0 213 Exclusive OR - Exactly one of the following processeswill start. The decisionor constraint effecting which processstarts is influencedby the control. Fan in junctions AsynchronousOR - One or more of the precedingprocesses complete, but not simultaneously. The conditions for completion are influenced by the control. Asynchronous OR - One or more of the preceding processeswin complete simultaneously. The conditions.for completion are influenced by the control. o MECHANISMS Mechanismsin IDEFo diagrams indicate the means by which a function is performed for example,a machine,computer, software or person instrumentalin completing the function. To correlate IDEFOand IDEF3 diagramsin an IDEFO-3 diagram it is necessaryto connect mechanismsidentified in the IIDEFodiagram with processesor junctions in the H)EF3 PFN. An example of the representationof mechanismsis shown in figure A2.4. In an IIDEFO-3 diagram mechanisms(solid lines) are shown entering through 'gates' in the bottom edge of the diagram to connect with either a fan out junction or a link. The diagram describesa situation where: RADAN nesting software is used to carry out the processes'Plan profile nests', 'Release profiling data to work centres', 'Releaseplate requirement data' and 'Releasevariations to 214 batch size'. The decision on which process or processeswill follow 'Plan profile nests' is taken at the asynchronousfan out junction by the mechanism- 'Planning engineer. Mechanismscan connectto the fbHowingjunction types: Fan out junctions & but Asynchronous AND - All the following processeswill start, not simultaneously, The conditions for start are influenced by the T mechanism following AsynchronousOR - One or more of the processeswill start, but not simultaneously. The decision effecting which processesstart is taken by the mechanism. following SynchronousOR - One or more of the processeswill start simultaneously. The decision effecting which processesstart is taken by the mechanism. following The Exclusive OR - Exactly one of the processeswill start. decisioneffecting which processstarts is taken by the mechanism. Fan in junctions Asynchronous OR - One or more of the preceding processeswin complete, but not simultaneously. The conditions for completion are influencedby the mechanism. 215 Asynchronous OR - One or more of the preceding processeswill complete simultaneously. The conditions for completion are influenced by the mechanism. A2.4 CONSTRUCTING IDEFO-3 DIAGRAMS The fbHowing stepsare usedto construct an IDEFO-3 diagram: 1. Establish the level in IDEFo decomposition appropriate for correlation with a process flow network. 2. In the context of the purpose of the model select appropriate IDEFo functions as scenariosfor processflow networks. 3. Identify the processes(UOBs) associatedwith each selectedscenario. Limit the number of UOBs to between 3 (to make the decomposition meaningful) and 20 (to limit the complexity of the diagram). If the number of UOBs is excessive consider either aggregating UOBs or further IDEFO decomposition before correlation. Diagram complexity can reach a point where too many relations detract from the clarity of the description 4. Develop a sequenceof processes 5. Develop the logical structure of processflow usingjunctions. 6. Correlate IDEFo inputs and outputs with the start and terminating processesin the PFN- Validate the correlation by identifying the sources of IDEFO outputs in terminating processes. 216 7. Correlate IDEFo controls with junctions and processes. Validate the correlation by identifying all IDEFo control connections. 8. If information concerning the influence of 'mechanisms' is necessaryfor the model purposeand viewpoint, correlate IDEFo mechanismswith junctions and processes. 9. If additional detail is required for the elementsof a PFN develop elaboration documents for UOBs,links andjunctions in the conventionalway. Diagram numbering., IDEFO-3 diagram numbers are all suffixed A03 followed by the EDEFonode (or scenario) number from which the IIDEFO-3diagram was decomposed. For example, an IDEFO-3 diagram numbered A03-223 indicates that 223 is the number of the node on the parent EDEFodiagram (diagram number A22) from which A03-223 was decomposed. The diagram developmentprocedure is inevitably iterative involving an analyst, a domain expert and others and typically severalinterview sessionsto develop common agreementon diagramdetail that matchesthe analystsdescription with the domainexperts knowledge. With a completeIDEFO-3 diagram a model can be developedby decomposingUOBs into greater levels of detail, if this is considerednecessary the conventional IDEF3 approachto decompositiondescribed by Mayer et al (1992a) is used. 217 A2.5 USING THE IDEFO-3 METHODOLOGY L;Affy Out IDEFO 'AS IS' functional analysis IýWcl'TO Using IDEFO-3 UseIDEF 0 'SOCnAri0e IDEFO-3 asbasis of -4 _ý'4 METHODOLOGY Courtiuction RwA=mmd TO BE construct IDEFO-3 diagram Analysis Figure A2.7 The IDEFO-3 methodology The IDEFO-3 methodologyis shown in figure A2.7. It consistsof sevenconsecutive phases. Phase 1 In phase I in the methodologyestablish a functionalAS IS IDEFO M- -ft a. - modelof the situation IDEFD-3 underreview. MEn4ODCXDW In this phase an analyst must establishthe purpose for the modelling effort as a commonlyagreed text definitionon the D)EFo contextdiagram. The text definitionalong with the B)EFO'Context' and'Viewpoint' diagramsare the top levelof an IDEFO-3model. 218 For the modelling effort to be effective the purpose and context for the intended model must map onto the manufacturing enterprisemodelling framework shown in figure A2.8 ensuring the purpose of modelling is compatible with cells 10,20,30 50,60,70 and is within the capability and scopeof the IDEFO-.3 method. To achieve this the top level of a model must exposethe focus and perspectiveof the modelling effort. ] b FSOCUS a C CONFIGURA TION BEHAVIOURAL QUANTITATIVE What, Where, Who How, When Flow Much, I low PERSPECTIVE Quickly, I low Manv I ENTERPRISE 10 50 90 Values, IDEFO Goals, Scope IDEFO-3 IDEFO-3 2STRUCTURAL 20 60 100 Functional IDEFO IDEFO-3 Relationships, IDEFO-3 Interfaces 3 PROCESS 30 IDEFO 70 110 The stagesinvolved IDEF3 IDEF3 in information and IDEFO-3 IDEFO-3 material flow 4DYNAMIC 40 80 120 The activation of information or object flow ENTERPRISE Product, Material, Constraints, Material Resourceconsumption, OPERA TION Resources,Locations, flow, Schedulesand Output, Performance Information. timing. Figure A2.8 The manufacturingenterprise modelling framework "IDEFO-3 - The framework indicaies' that 'can model; Configuration to answer interrogatives such as what?, who? where?, i. e. the objects, material, products, information, data and resourcesused by an enterprisetogether with their relationships. Behaviour to describe how manufacturing is conducted, for example to represent the relative timing of events, the logic of processes,what initiates, enables,constrains and inhibits manufacture. 219 It also indicatesthat IDEFO-3 can represent:An Enterprise perspective, a high level view that considersthe manufacturing scope of an enterprise,the products, its overall goals, operation and values and the manufacturing strategy that it uses to achieveits goals. A Structural perspective that examines the relationships between the elements of an enterprisein a static functional view to describesthe interfaces, links and relationships that exist. A Process perspectivethat looks at the actions that take place to changethe state of material or information and the sequenceof events and the cause and effect of events. Phase one is completed by developing an AS-IS 11DEFomodel using the model building techniquesdescribed by the IDEFo manual(US Air Force, 1981 a). Phase 2 and 3 In phase 2 selected IDEFo functions become scenarios for decomposition into IDEFO-3 diagrams. This stageis reachedwhen functional METHODOLOGY descriptions become sequential narratives or when knowledge of aspects of the systems behaviour is necessaryto achieve the modelling objectives'. In phase 3 IDEFO-3 diagrams are completed using the rules and techniques described in sectionsA2.3 of this user guide and validated through the 'Author Reader Cycle' in the usual way (US Air Force, 1981 a). 220 Phase 4 and 5 - 11 c..,..lDIg _____J, MUß) VEF" METHODOLLM In phase4 the AS IS IDEFO-3 model is analysed '-. u by involved in > ------all those the modelling effort. Gaps and inconsistencies between the perceptionsof those involved are resolved. An outcome of this phaseis agreementon the AS IS model as a common understandingby both domain expertsand analysts. Phasefive involves recommendingthe changesnecessary to achievethe modelling goals this becomes the context and viewpoint of the TO BE IIDEFO-3 model to be reviewed and changedin light of the problems,inconsistencies or generalfindings from analysisof the AS IS model. Phase 6 In phase 6 the TO BE IDEFO-3 model is developed, firstly defining a purpose context and viewpoint and D3EFodiagrams to describe the planned state of system incorporating the recommendationsagreed in phase5. The TO BE IIDEFO-3model is createdand correlatedusing scenarios identified as important in the higherlevel IDEFodiagrams. The modelis then validatedthrough the authorreader cycle. 221 Phase 7 ý11 Implementation takes place in phase 7. -S 151713 The nature of this phase is wholly MErHODOLOGY dependent on the project and on the ý a. ' implementation procedures of the enterpriseinvolved. / 222 APPENDIX 3 A BTUEEFREVIEW OF THE IUDEF METHODS 223 CONTENTS A3.1 IDEFo 224 ...... A3.2 IDEF3 226 ...... A3.3 IDEFO IDEF, MODELLING TOOLS 227 and x ...... A3.4 IDEF, 231 ...... A3.5 IDEFix 235 ...... A3.6 IDEF2...... 241 A3.7 IDEF4 247 ...... A3.1 EDEFo The complete IIDEFOmethod is defined by the 'Architects Manual' (Soff ech Inc., 1980) it includes guidance for modelling, together with rules for model syntax, diagram and model format, text presentation as well as a structured model validation, document control proceduresand interview techniques. A key concept of IDEFOmodelling is the definition of a 'Context' and the modeller's 'Viewpoint' to establishan explicit common understandingof the boundary and aspect of the systembeing modelled. The first step in IIDEFomodelling is thus concernedwith establishingthe objectives of the modelling effort from which a context and viewpoint can evolve. An BDEFOmodel consists of a hierarchy of related diagrams that representsan aspect of a manufacturingsystem (the method is also widely used outside manufacturing,the focus of this work is howevermanufacturing enterprises). Each diagramis basedon a diagonalrow of 224 boxes (normally between three and six boxes on each diagram) connectedby a network of arrows. Boxes representactivities or functions of a systemand a text block inside a box is an active verb phrasedescribing an activity. Arrows representthe relationshipbetween activities in terms of the information or objects used, produced or required by activities. Arrows entering the left side of a box are inputs (I) to the activity, arrows entering the top of a box are controls (C) on the activity and arrows leaving the right side of a box are outputs (0) as a result of the activity. Finally a mechanismM is a person, systemor device associatedwith carrying out the activity and is shown as an arrow entering the base of a box. This arrow structure is termed the ICOM structure. Each diagram is referred to by its 'node number' that defineswhere it Les in the hierarchy of a model. An exampleof IIDEFoactivity diagram format is shown in figures 5.7 to 5.13. A node index and node tree can provide a convenient way of showing the relationshipbetween all the diagramsin a model, examplesare shown in figures 7.12 and 7.13, and a data dictionary can be used to summarise all sources and destinationsof objects,information or data (Colquhoun et al, 1989). An IDEFo model comprises of a set of related activity diagrams presentedin node number order that representthe subject under scrutiny. The perspectivetaken by the author of the subject and the scope of the model is defined by the 'context' (node A-0) and 'viewpoint' (node AO) diagrams at the apex of the hierarchy. The depth of the model hierarchy (the number of diagramsbelow the viewpoint) is determinedby the amount of detail required by the analyst. Supporting the basic principles of the method is the IDEFo forms and proceduresguide (Ross et al. 1980) and the ICAM library maintenanceand distribution procedures (US Air Force, 1981e) The procedures,based on the 'Author/Reader cycle' provide a structured meansof controlling, documenting and validating the model building process. The final model is thus the shared understandingof a team of people producing a common representation of the subjectunder review. 225 A3.2 IIDEF3 A completeguide to the IIDEF3 method is provided by Mayer et al (I 992a), what follows is a brief description of the main characteristics of the method. IIDEF3 is used to describe a systemby capturing a representationof what a system does, rather than to predict what a systemwill do. It uses ProcessFlow Network (PFN) diagramsand Object State Transition Network (OSTN) diagrams. The PFN diagram represents Units Of Behaviour (UOBs) (where a UOB could be: a process,a function, an activity, an operation, a decision, an action, or an event) connectedby their temporal, causaland logical relationships. An OSTN diagram describesthe various states of objects used in processesand relates those states in terms of the processesthat causeobjects to changestate. Obtain Set 400 1 1--1! 4 P_M. P41 up IA ýF T. I. Run4OOpnw 'M, ý-4 numnal Mal -4-14 1.2 1bWPOd A LG =*I Puk prod= M.W. 1 PFOCM Figure AM A ProcessFlow Network - Run Cell. The IIDEF3 method is focused on capturing the knowledge of the experts involved in the system and uses graphics and text to describe the observations,beliefs and statementsof participantsin the system Figure AM is an exampleof a PFN that describesthe operation of a manufacturing cell. The large rectangles labelled in the bottom left hand comer are UOBs describing the processesinvolved in the 'scenario' of 'Run Cell' for example 'Obtain material (2 1.1) and 'Set up 400 press' (1.1) are processesthat occur sequentially(indicated by the precedence 226 Link L5). Smaller rectangleslabelled with a 'J' are Junctions representingthe divergence or convergenceof sequencelogic. The box labelled JI in figure A3.1, for instance, representsan 'asynchronousAND' situation using a fan out junction, where the processes 'Run 400 press and Return material', 'Inspect & pack product', and 'Monitor process' all occur concurrently. 'Asynchronous' indicatesthat they do not necessarilystart at the same time and 'AND' indicatesthat they all do occur. In common with other modelling methods the principle of decomposition can be applied to 'explode' a UOB into lower levels of detail using formal syntax to structure the process. In Figure AM those UOBs depicted with a shadowed rectangle have been decomposedto a lower level. Junctions and UOB's are connectedby 'Links' depicted by arrows. Links can representprecedence, object flow or a user defined relationship. Each UOB, Junction and Link can have a structured 'Elaboration document' to describethe elementin more detail. A3.3 IDEFO and IDEF, x MODELLING TOOLS Sincethe IDEFO,I and Ix methodshave been in the public domaina rangeof softwaretools havebeen developed,many of which are marketedas integrated software suites: Wisdom SystemsInc. (1300 Iroquois Avenue, Naperville, Illinois) was founded by Dennis Wisnosky in 1986 (Jenks, 1993). Wisnosky was program managerfor the ICAM project and is credited along with Ross with coining the IDEF acronym,(Wisnosky, 1979). The company beganby marketing the first commerciallyavailable IDEF tools - IDEFine-O and IDEFine-1 I. Using their current product Wisdom systemsclaim to have establishedthe term Computer- Aided ProcessRe-engineering (Wisdom Systems,1995) basedon an 'integrated set of tools' consistingof five modules. I IDEFine-Oand IDEFine-I were a product of Sophides;International, Churchillstraat35, Bamavcld, Netherlands 227 I. Minerva -A BPR methodologyto integrate the modules 2. ProcessWorks - IDEFOmodelling 3. Data Works - IDEF, x modelling 4. Wisdom model repository -A databasefor model management 5. Other Wisdom analysistools provide a meansof analysingIDEFo models: 9 TimeWIZard, time line analysis(Gantt chart) e CostWIZard, Activity Based Costing * QualWIZard,Quality Function Deployment * SimWizard,for export of functional modelsto ProModel or CACI SIMprocess discreteevent simulation tools. Triune Software Inc. (2900 PresidentialDrive, suite 240, Fairborn, Ohio), founded in 1988 produce the Automated BusinessLogic EngineeringProcess Modeller (ABLE PM) tool, it is an IDEFo modelling tool that claims to 'help manufacturersretailers or service providers re- engineertheir way toward enhancedproductivity' (Triune Software, 1993). No links with other tools are claimedand no IDEFIx modelling capability is provided. Coe-TruemanTechnologies Inc., (1321 Duke Street, Suite 301, Alexandria, Virginia) formed in 1985 produce COSMO an 'Interactive Enterprise engineeringand CASE tool' consistingof five modules. 1. COSMO-SP- a strategicplanning and model managementfacility 2. COSMO-0 - an IDEFOmodelling tool 3. COSMO- ABC Activity Based Costing using EDEFomodels 4. COSMO-IX - IIDEF,x modelling 5. COSMO-PRO a 'User view' generator using SQL to produce databasequeries and reports. The module uses COSMO-lX IIDEF,x models to construct normalised databases. 229 Knowledge Based SystemsInc. (KB SI), (1500 University Drive East, College Station, Texas) founded in 1988 by Richard J Mayer (Mayer 1993a) produce an 'intelligent workbench' consistingof six modules. 1. AIO WIN - IDEFo modeffing 2. Pro ABC - Activity basedcosting using AIO WIN IIDEFomodels 3. ProCap - Processmodelling using IDEF3 4. ProSim - Processmodeffing (IDEF3) with an interface to AT&T ISTEL's WITNESS discreteevent simulation software tool. S. ProCost - Activity basedcosting using ProSirn IDEF3 models 6. SmartER - Information and Data modelling using IDEFI and IDEFx Logifforks Inc. (214 CamegieCentre, Princeton, New Jersey)is describedby Frank (1993) as ta relatively inexpensiveCASE tool' consistingof four modules. 1. BPwin IDEFo modeffing - , 2. ERwin/ERX - IDEFIx modeffing 3. ERwin/DBF - design and reverse-engineeringfor databases(dBASE, FoxPro and Clipper) froni/to ERwin/ERX IDEFix models. 4. ERwin/SQL- GeneratesSQL schema'sand indexes from ERwin/ERXIDEFIx models for SQL databaseappfications. Texas instruments (AIM Division, 6550 Chase Oaks Boulevard, Plano, Texas) produce the Business Design Facility (BDF) which is a unified CASE tool that incorporates several diagramming methods, Process outline diagrams,Process maps, Process chains and IDEFo diagramstogether with IDEFIx diagrams. 229 MetaSoftware (125 Cambridge Park Drive, Cambridge, Massachusetts)founded in 1985 produce a combined IDEFoADEFjx tool called Design/IDEF. The company claim that it is the only IDEFOtool that has an integral Activity Based Costing facility. In 1988 the company begandevelopment of Design/CPN (Coloured Petri-Net) claimed by its designers(Albrecht et al, 1992) to be the first software package to support hierarchical coloured Petri-nets. The tool is linked to a Design/IDEF IDEFo model to build a CP Net graph. IDEF software tools from the various vendors are not integrated although some work is being carried out to provide a common software tools environment. In the USA the National Centre for Manufacturing Science(NCMS), for example,have establishedan IDEF Model Repository System designed to facilitate accessto all the various modelling tools (Cauthorn, 1993). All the modellingtools discussedoriginated in the USA in the mid to late 1980sand the advertisingliterature relating to the varioussystems emphasises the stronginfluence of US governmentrequirements (see section 2.3) andindicates that the majormarket in companies involved in US governmentcontracts. In the UK the only tool actively promoted is MetaSoftware'sDesign/IDEF and approximatelyone hundredand twenty five copieshave beensold in the UK over the last five years(see section 3.5). Logic Works have(in 1994) appointedAdmiral Software(Camberley, Surrey) as their UK agent,sales of this tool are unknown. The approachadopted by most tool vendors has been to provide IDEFo modellingas part of a CASEtool environmentand reflects the emphasison IS applicationof the method. The recentrise of BusinessProcess Re-engineering (BPR) in the USA as a paradigmfor the managementof change(Drucker, 1991, Hammer, 1990) has provided fresh emphasison the needto understandand design processes as the basisof commercial,service and manufacturingenterprises. Which in turn has provideda new marketingimpetus for IDEF tools underthe bannerof BPR tools. 230 This is to some extent reflected in the change of name of the US IIDEF User GrOUP2, which was formerly a forum for IIDEF practitioners and tool vendors, to The Society for Enterprise Engineering in 1995 to 'promote and expand the practice of analytical techniques, specificallyIDEF, in BPIV (Preston, 1995). This research in connection with IDEF tools led to a worldng relationship with the UK vendor for MetaSoftware's Design/IDEF and to the company's support for the survey of usersdiscussed in section3.5. 3.4 11DEF, The need for the IDEF, method 'became clear as the difficulty in designing integrated manufacturing systems became more and more apparent' (US Air Force, 1981b). The method was designedto produce an information requirementsmodel before designing and building an information system. It is capableof capturing and graphically representingwhat information is necessaryto manageand operate an enterpriseand can include, not only the information necessaryfor computer systems, but also physical entities such as material, products, and equipment. It was designed to analyse and describe information resource managementrequirements rather than to design databasestructure. A model can describe the logical relationshipsin information flow and the rules controlling the managementand storage of information and can be used to provide the knowledge and insight necessaryto identify problemsand build a strategyfor information management. IDEF, modellingprinciples: IDEF UsersGroup, Kettering, Ohio, USA, founded1989 (Preston, 1990) 231 The graphical conventionsused can represent; objects, the physical or abstract relationships between objects, information concerning objects and the data structure used to deal with information. The basic building blocks of an IDEF, model are 'entity classes'comprised of groups of entities. An 'entity' for example could be real, such as a 'design manager', a 'solenoid valve' or a concept such as 'manufacturing problem'. An entity has properties or attributes, for the solenoid valve, 'part number', 'description, 'cost' and 'supplier name' could be the information or attributes associatedwith a particular solenoidvalve to identify it uniquely and differentiate it from another entities. Eachentity is an individualmember of an 'entity class'which representsthe informationthat a group of entitieshave in common. In the caseof a solenoidvalve for exampleseveral couldbe usedby an enterprise,single acting, spring to centre,three position etc.. An entity classrepresents the informationknown about the similar propertiesof the collection of entities,for example,'supplier', 'port size', 'pressurerange' etc.. The first stagein building an IIDEFImodel is to build an entity classglossary or pool in the form of a fist or table. Each entity classis an entryin a tableand eachentry hasan associatedentity classdefinition on a separateform. The secondstage is to establishthe 'relationclasses' between entity classes.A relationclass describesthe mannerin which membersof an entity classrelate to membersof anotherentity class(or entity classes).For examplethe entity class'solenoid valves' are 'used on' (the relationclass) another entity class'hydraulic power pack'. IDEF, usesa relationalmatrix simplyto indicateif thereis a relationshipbetween entity classes. To describethe cardinalityand dependency between entity classesthe graphicalconventions shownin figuresA3.2. andA3.3 areused. FigureA3.3 describesthe following relationship: Any 'solenoidvalve' entity may be used on zero, one or many (0,1, N integer values) 'Hydraulic power pack' entities. Each 'Hydraulic power pack' entity usesprecisely one solenoidvalve. The diagramalso definesthat no other entity classesrelate to 'solenoid valve' or 'Hydraulic power pack' entity classesand that no other relation class exists betweenthe two entity classes. 232 1100,1, N 1,N Oll Oll 1< Oil Figure A3.2 IDEF, graphic notations to specify relation classes Referencenumber Entitycim Wdon class Figure A3.3 A simpleIIDEFI entity classdiagram 233 fart No. Key (underlined) Suppjief AtUbutsclass Type Nm4c&y I &WbuteCie Solenoidvalve Awaaft No Conq)oundkey Ojetomer attffatft Cku" p- plm* Figure A3.4 A simpleIDEFI entity classdiagram showing attribute classes The third stageof the method involves defining the attributes of entity classesin order to be able to distinguishindividual membersof the sameentity class. 'Key' and 'non-key attribute' classesare used. Key attribute classescan be single or compound (two or more attribute classes). Figure A3.4 illustrates single and compoundkey attributes. Finally phase four involves the assigmnentof non-key attribute classesto entity classesto completethe description of entity classes. Alongside the principles of the method the IDEF, methodology uses the Author/Reader cycle for model building and validation in commonwith EDEFo(US Air Force, 1981 a, b). IDEF, is a method for documentinginformation requirementsand structure as a foundation for databasedesign it is a knowledge gathering method designed as an 'AS IS' method rather that databasedesign method. For the task of databasedesign the IDEFIx method was developedusing the experiencegained in the developmentof IDEFI. 234 A3.5 IUDEFx EDEF,x is a databasedesign method that is independent of data storage (i. e. the target database)and the user interface and is one of many data modelling approachesbased on the work of Chen (1976). It was designedas a method to design relational databasesand is 'not particularly suited to serve as an AS IS analysistool' (Mayer, 1990b) and is not considered useful for object-orientedsystems. The method is used to develop a conceptualschema, that is, the logical structure of data independent of the application of the data (the external schema)and how it is accessedand stored (the internal schema). IDMX modelling syntax and semantics: An outline of the key IDEFIx modelling concepts; entities, relationships and attributes, is given in this section. The details are based on the original US Air Force (1986) IDEFIx report, Mayer (I 990b) and the work of Bruce (1992). (i) Entities. fundamental IDEF, and Ix terminology is similar however there are differences in the conceptsemployed. In IDEFjx an 'entity' is a set of real or abstractitems of interest such as components, staff, concepts or events sharing common attributes. Individual members of sets(single occurrencesof an entity) are 'entity instances'. 235 Entityname PROJECT11ýý Project D. An identifier Ptimarykey Independententity attributes Entityname CUSTOMER-J---- CustomerNo. -` 7--", Primatykey An identifier attributes dependententity Figure A3.5 Graphicalrepresentation of DDEFIxentities An entity is termed 'independent' if each entity instance can be uniquely identified without knowing its relationship with other entities. An entity is 'dependent' if its relationship with other entities needsto be known to uniquely identify its entity instances. (ii) Relationships. Relationshipsbetween entities are based on the parent-childconcept where each instance of an entity, referredto as the 'parent entity' is relatedto zero, one or more instancesof a secondentity calledthe 'child entity' and eachinstance of the child entity is relatedto one specificinstance of the parententity. The instanceof the child entity can only exist if its associatedparent entity instance exists. 236 PARENT A parent entity instance may have zero, one or more associatedclild entity instances(one-to-zero-or-more). 41 CHILD P Each parent entity instancemust have at least one or more associatedchild entity instances. 41 Z Each parent entity instancecan have none or at most one associatedchild instance. 41 Eachparent entity instanceis associatedwith someexact number of child entity instances. Figure A3.6 IDEFIx relationshipcardinality The relationshipcardinalities detailed in figure A3.6 can be describedusing IDEF, x. in figure A3.6 the solid he representsthe identifying relationship between parent and child entities, when a relationship exists the child entity is always identifier-dependentand representedby a box with rounded comers. In such a casethe primary key attributes of the parent entity are also inherited primary key attributes of the child entity. In an identifyingrelationship a parententity is identifier-independentunless it is also a child entityin a relationshipwith otherentities, in which caseparent and child entitiesare identifier- dependent. 237 A non-identifying relationship is described by a dashedline. In which caseboth parent and child entities are identifier-independentunless they are in some other relationship which is an identifying relationship. PROJECT Project I.D. Parent menanerappenaenr Non idenfifying entWes relatfonship ORDER Order no. Child Figure A3.7 An IDEF, x non-identifying relationship In non-identifying relationshipsan additional discriminator is used in the presenceor absence of a diamond at the end of a relationship symbol. A one-to-somethingrelationship is denoted without the diamond as the examplesin figure A3.6. With the diamond it becomeszero-or- one-to-something,an exampleis shown in figure A3.8. Zero-or-oneto something 16 Figure A3.8 A zero-or-one to somethingIDEFIx non-identifying relationship 238 Categorisationis used in IDEFIx to group entities hierarchically when they share common characteristics. An example of the graphical symbols used is given in figure A3.8. A discriminator, Valve "e, is used in the example, each instance of the parent entity Valve is associatedwith one instance of one of the child entities, Manual valve, Pneumatic valve or Solenoid valve. Valve Parent (generalisatfon) endfy bwww4w*wxwas &, Imr (of IdertMer-dependemdepending on odw fohd-"PS) Valve type Solenoid I Pneumatic valve Manualvalve vAip ory EntWes Ahwyeklandhw-dependent ondbas Figure A3.9 IDEFIx categorisationrelationsWps (iii) Attributes. An attribute is a set of certain properties of an entity, and an attribute instance is a individual specific characteristicof an memberof the set. Attributes can be: Primary key - an attribute that has been selectedas a unique identifier for the entity. Non-primary - an attribute of an entity but not a primary key. Candidatekey - An attribute (or group of attributes) that could be selected as a primary key. Alternate key -A candidate key not selected as the primary key. Foreign key - is an inherited attribute in a parent-child relationship, where the primary key attribute of the parent entity is also an attribute of a child entity. Figure A3.10 is 239 an extract from a data model for Engineering Data Control (EDC) in capital plant manufacture Wustratingthe representationof attributes. keys PROJECT CUSTOMER Projid: Integer not null Custnolnteger not null PPS-Ordemo: Char (5) (FKj Name: Char (18) PPS- Bidno: Char (6) (FK) Address: Char (50) Projno: Integer Not null Country: Char (18) Custno: Integer Not null (FK ProLcomponents I- (Fig = Foreign key Char (n) or nature of ýProjid:Integer (FK) not null Integer not null attribute IChaptno: Integer not null (FK) PPS Ordemo: Char (5) (Fiq 1 Supýller Char (10) Figure A3.10 Extract from 'EngineeringData Control (EDC) in capital plant manufacture' (source:Hubel and Colquhoun,1996) The syntax and semantics of IDEFIx provide a phased methodology to designing an information system. As part of the methodology the IDEFo forms and procedures guide (Ross el al. 1980) and the ICAM library maintenanceand distribution procedures (US Air Force, 1981e) (in common with other IDEF methods) provides rigour to validation, documentationand model development. The method is widely recognisedin the USA, the National Institute of Standardsand Technology (NIST) has funded efforts to establishIDEFIx as a Federal Information ProcessingStandard (FIPS) and section2.6 indicatesthe level of softwaresupport available for the method. Thereis, however,no evidencethat the methodis widely usedin the UK or Europe. In the contextof this researchthe methodserves as an exampleof entity-relationship IS modellingand as a complementarymethod to 11DEFOand IDEF3. Much of the modelling 240 effort in manufacturing enterprise systems is ultimately aimed at introducing modifying or implementing software solutions. Any functional or process modelling approachesmust be compatiblewith the more exhaustivedetail requirementsif IS modelling which is an argument for using methodssuch as IDEF in preferenceto methods such as GRAI and Checkland. A3.6 IIDEF2 IDEF2 is the ICAM dynamicsmodelling methodology, 'designedto describethe time varying behaviour of manufacturingsystems in such a way that the descriptionscan be analysedusing computer simulation to generate measuresof manufacturing system performance' (US Air Force, 1981c). It was specifically designed to represent 'a wide class of manufacturing systems' production lines, group technology cells, software systemsor procedural systems such as shop order releasesystems. The method is designedas a descriptive and an analysis tool and Pritsker and AssociateVpropose that an IDEF2 model can be employedin five ways: As an explanatorydevice to define a systemor problem As a documentationmedium As a communicationsvehicle to determinesystem elements components or issues As a designassessor to synthesiseand evaluateproposed solutions to problems As a predictor to forecast and aid in planning future developments The rationale behind developmentof the method was that the ICAM methods under developmentin the late 1970's (IDEFo and IDEF,) could not describethe dynamicsof functionsor information. In additionto the aspectsof IDEFoand IIDEFImodels that could vary overtime, therewere other aspects of manufacturingsuch as queues and decisions which demandeda modellingmethod. The needto evaluatespecific measures of manufacturing systemperformance such as; throughput, the abilityto meetdeadlines, resource utilisation, in- processinventory, was alsoconsidered an objectivein the developmentof IDEF2. 'Pritsker andAssociates, The companywith responsibilityfor developmentof the method,Milner, PLJ., Pritskcr,A. A. B., andIppolito, J. F. werespecifically responsible for developingthe syntaxand semantics associatedwith EDEF2. 241 IDEF2 modellingprinciples: This section is a summaryof the principles of IDEF2 modelling, the major referencework is the Architects Manual (US Air Force, 1981c) which provides a detailed account of syntax, semanticsand methodology. An IDEF2 model comprisesfour sub-models: (i) An entity flow sub-model that describes entities, such as material or information, the sequenceof processesthey undergo and what alternative processesare possible. Arrows representactivities and nodes representthe start or end of activities where queues,decisions, and events occur. Figure AMI is a flow network showing the symbolsused. Activities or processesare shown as arrows and labelled, Queue nodes are shown using 'accumulate' symbolsand resourcerequired and resourcereleased states are shown using split box 'assign' and 'allocate' symbolstogether with labels. The 'boundary' of the systemunder investigation is shown using a split circle with a 'type' label in the upper half and a location number in the lower half. Boundary types can be START, STOP or GO TO (another location). Enthyflow Boundarysymbol (indicates Queue where Acttvides an enthyenters or exits) ýnodes --\, CLEM K I I OPEMTOR NSPECTOR1,1INSPECMR !L Typeof resource required for the 'CLEAN'acdvity Quantityof resource required symbol resource released Figure A3.11 An IDEF2 entity flow diagram 242 Entity flows in a network can have different routes and are shown by diverging arrows after aselector' symbols,an exampleis shown in figure A3.12. A complete diagram or 'entity flow network' may contain severalbranches to representrouting decisions,entity selection, entity matching or branchingbased on probabilities or conditional logic for eachflow path. Other symbolsused in IIDEF2are the 'match node' used to describea condition where entities in a queueare matchedon the specificvalue of an attribute and the 'assemblynode' to describethe situationwhere several entities from differentqueues are assembledinto a single entity. MLUNO ) P""RT QUEUE DRLUNG A branch for an entfty to take An acdvfty for an enthy to wVags In Resources to perform acWbes An endty from parallel queues A queue to hold ottles Figure A3.12 Alternative entity flows 243 Rwoufcodispas; don ach as: ftraquests? MACHNE Resouma *9 be "AV& In lime units? Is ft number In OWLIE a gAw Action WW - poWbIs vahm? NO YES actions Aflocam Five ANYREOLIESM Preen%% Enor Resource We ALLOMTE Abdo b which macNne /a NUnbarof sibcated Awwrces Figure A3.13 A resourcedisposition tree (AdaptedfromUS Air Force,1981c) (H) A resourcedisposition sub-model used to describethe stateof resources.A hierarchical tree structureusing QUESTIONSand ACTIONS describesthe dispositionof a resource when an activity is completed. Figure A3.13 shows a resourcedisposition tree for a 'machine' resourcethat processesparts from a queue, when a part is completedthe dispositionof the machineis established. The 'any requests'box polls the machineto seeif any parts require processing. If the answer is no the resourcecan be freed, if the answer is yes the resource 'machine' is allocated to the machine queue. Four possible actions can be initiated using a resource disposition tree; allocate, free, pre-empt and error. (iii) A systemcontrol sub-modeldescribing the actionsof activitiesthat control the flow of entities. The controldescribed in a SystemControl Network is initiatedby events,milestones or decisionswhere the statusof the systemmay change. Figure A3.14 is a simpleexample of a SystemControl Network describingthe arrival of entitiesinto a system. The CREATE node describesthe generationof entities,the SELECTORnode feedsentities to boundary nodesin entity flow networks. The control over which entity goes to which entity flow network is definedby the branchlabels. In this casethree controls are defined;that the 244 number of entities is greater than 25, that the number of entities in QUEUE I is less than 12 and that the current time is lessthan 1000 units. Cor4rols ow onfity flow twough branchm NUU GT.25 'Nods lo numbers fof r NUM OUE1 LT.1 2 S13 nodes in enbty Ilow netwoft ý010 node E Ltl Cromienode Figure A3.14 A SystemControl Network (Source:US Air Force,1981c) Other nodesthat are usedin SystemControl Networks includethe ENTER node which is usedto initiate an eventoccurrence and the ACTIVATE/DEACTIVATE nodewhich is used to describeresources available or not availablefor use. In additioncontrol network nodes include ALTER (resourcecapacity), DETECT(variable values) and PREEMPT (resource use). (iv) A facility sub-modelwhich describesthe relationshipsbetween equipment or resources used by the system. The model could include physical, procedural or intellectual resources. This aspect of IDEF2 modelling uses sketchesof the relative locations of system elementsin the initial modelling phase to provide a basic understandingof the system under review in terms of layout and production flow. The symbols used to describe elements have been adapted from conventional computer Ilow chart symbolsbecause 'no standard methods are available, many plants use different layout techniques. However the emphasis when constructing a layout is usually on identifying the flow of materials through the facility' (US Air Force, 1981c). The sketchesused in this phase of modelling are not consideredto be formal aspects of a complete IDEF2 model but as a structured means of gathering information. 245 In additionto the four diagrammingtechniques supporting documents are used to specifythe characteristicsof the elementsin diagrams. A QUEUE for examplecould be characterised by; an identifier, a description,the ranking rule to extract from the queue,the maximum capacity,the initial numberin the queueetc. The formsused to defineattributes and values to diagramelements are: FacilityComponent Attribute Definition Entity Definition Entity Flow NodeDefinition Initial Entity Disposition(the initial statesof queuesand entities in activities) Initial ResourceDisposition (the quantityand initial dispositionof eachresource) Initial ContinuousSystem Definition SystemControl Node AttributeDefinition The IIDEF2 methodology also includes details of the control of 'Kits' (diagrams, text, glossariesand backgroundinformation for review and comment)through the ICAM library maintenanceand distribution procedurescommon to other IDEF methods. During this researchthe only complete IDEF2 model encounteredis a detailed model (280 pages) of 'A Sheet Metal Centre Sub-system' produced by Soffech Inc. under USAF contract F33615-78-C-5158(U S Air Force, 1981f) to demonstratethe usability of IDEF2 on a real world problem and 'to build a model which can be analysedby IDEF2 software when it becomesavailable'. It is clear that commercialIDEF2 software did not becomeavailable, that the US Air Force contract (Pritsker and Associates,see section 2.3.2) to develop the method stopped short of funding software developmentand the method has not been widely adopted. Bondi and Pritsker (1985), however, provide an account of a prototype system for the analysisof IDEF2 networks called NET_TRANS (Net Translator). The systemautomatically determines the applicability of either the Graphical Evaluation and Review Technique (GERTE) (Pritsker, 1974) or queuingnetwork theory for the analysisof an IDEF2 models. SLAME is one of the four most widely used general purpose simulation languagesin the USA (Law and Kelton, 1991), it was developedby Pritsker and Associates(Pritsker, 1986) 246 from their earlier SLAM simulation software, Ross claims (1994) that IIDEF2 was the foundation of SLAMIE[although no acknowledgementof this is given by Pritsker in his work. SLAMII is one of a group of simulation languagesthat uses graphical networks as the basis of description (Yancey, 1985). The symbols used in a SLAME network, nodes, queues, assignments,etc. are analogousto those in a IDEF2 entity flow network although SLAMII carries more detailed parametervalues in diagramsand the method is focused on 'statements models' (text statementsof the sequenceof entity flow through a network) for input to a computer. Despite the lack of development of IDEF2 it still played a part in the evolutionary developmentof simulation for manufacturing systemsand more importantly for this research, had someinfluence on the developmentof IDEF3. The philosophy behind both methodsis to provide a meansof describing the behaviour of a systemin such a way that the description can be analysedusing a computer systemto evaluateperformance. Both methods emphasise the use of the graphical aspectsof a model as a vehicle for communicationand both methods recognisethe needto representthe 'time' or temporal of manufacturing. A3.7 IUDEF4 IIDEF4 (Mayer et al, 1992b) is a design method that supports an object-oriented approachto software design in contrast to IDDEFIxwhich is based on structured analysis and functional decomposition. The object-oriented approach is an emerging philosophy now widely acceptedas a meansof thinking abstractly about software design problems independentof a software implementationlanguage. Object-orientedmodelling uses the concept of modelling real world 'objects' that combine data structure and behaviour in a single element and then utilising the model to design a systemaround the objects, independentof the languageto be used for software implementation. Object-orientedapproaches describe entities through data abstraction encapsulation and inheritance. Using the approachthe systemunder review is seenas a collection of objects which are encapsulationsof data whose functions are defined by the messagesto which they respond. Through encapsulation,adding or deleting objects 247 doesnot influenceother objects. Objects also belong to classesor sub classesand they inherit the behaviour and structure of their super-class. New objects can be added easily by inheriting the characteristicsof their class. Many authors (Coad and Yourdon, 1990,1991, Rumbaugh et al, 1991, Agresti, 1986, Somerville, 1992) advocate advantagesfor object oriented software design and analysisover traditional functional approaches; software re-use and flexibility due to modular design. software maintainability due to transparent system design, relative ease of program modification. the clear relationshipbetween real world entities and systemdesign. The real world tends to be perceivedas physical entities rather than functions. Severalmodelling methods(Booch, 1991, Rumbaughet al, 1991, Firesmith, 1993, Shlaerand Mellor, 1988, Graham, 1991, Coad and Yourdon, 1990) have been developedto support the concept, amongst them the IIDEF4 method. In the original version of IIDER (Mayer et al, 1992b) objects are modelled using a hierarchy of related diagrams,figure A3.15 shows the organisationof a completemodel. Revisiontwo of the methodwas a substantialre-design using important features of successful object oriented designsand modelling conventionsto 'leverage software components, chent/servertechnology and successfulobject-oriented design technology' (Keen, and Schafrik1995). 248 Type InstantiatiOn diagrams diagrams Protocol Cum dagrams sub-model Inherltance CUsslnvarient-j IDER diagrams data sheets model Dispatchmapping Method ý Contract wonomy Method data sheets sub-model Client diagrams Figure A3.15 The structure of an IDEN model (Source:Mayer et al, 1992b) Revision two-IDEF4 is very similar to Rumbaugh'sObject Method Technique(OMT) and the Object Oriented Analysis and design technique (OOA/OOD) of Shlaer and Mellor with three crucialdifferences: It is specificallydesigned to interface and re-use information generatedin IDEFO,I1DEFjx and IDEF3 models. * It allowstracking of the statusof designartefacts from domainobject through transition to designspecification. * It includesa designrationale component. IDER revision two (Keen, and Schafrik 1995) is based on three models: a Static model (SM), Dynamic Model (DM) and a Behaviour Model (BM) together with a design rationale component: (TableA3.1 Givesexamples of someof the termsused in IDEF4 revisiontwo) 249 An SM defines time invariant relationships between objects such as inheritance, object classification,operations that canbe performedon an object,attributes describing objects and structural relations. A DM defines communications between objects using a graphical representation of the messagesrelayed between objects and the eventswhich causean object to implementa message. A BM is a diagramthat describesthe implementationof messagesby methods. IDEF4 Term IDEF4 Examples symbol Application Bank domain Feature Aspects of the 'Bank' to be modelled Object 0 Account, 0 Employee Attributes A Name, A Address Methods M Open, M Close Relation R Emp/Acc Event E Audit Table AM Examples of IDEF4 terms (Adaptedfrom Keen, 1995) Design rationale capture is accomplishedby providing specificationsfor each IDEF4 arlefact (figure A3.16 shows the evolution of IDEF4 arlefacts). Figure A3.16 also illustrates the relationship between artefact evolution and design status. The essence of the IDEF4 methodology is that for eachstage in design status SMs, DMs and BMs are produced. 250 A sottware life cycle from conception to system Implementationin an Object -oriented language Application domain Specification Software Design artefacts artefacts artefacts artefacts Anobject abstracted tromto application domainag. A Bank, anAircraft RequirementsanaPysis for theapplication domain becomesa specification for DesignarWfacts Softwarespecifications evc*.fed from design artefactsto become Specificationartefacts; softwaredesign evolveto become specification softwarepackages containIng artafacts relatedprocedures and data It Designstatus of software DONWN TRANSITION SPECIFICATION Figure A3.16 The evolution of IDER artefacts The links between MER version two and other IDEF methods are of particular interest to this research.Although no explicit indicationis givenof the potentialinterfaces it is evident that they are in the areasindicated in table A3.2. IDEF4 version 2 IDEFO IDEF3 Method Activity UOB Features Viewpoint Scenario Event UOB & Junction Object & Attributes ICOM Table A3.2 Potential relationshipsbetween IDEF4 version 2 and other IIDEF methods 251 IIDEF4 version 2 is evidently in the development stage, little evidence (Zhao, Baines and Colquhoun, 1995) has been found of its application outside the work of the developers, KBSI, and no commercial IDEF4 software tool is available (KBSI are currently (1995) developingan IIDEF4 tool called A141m). There are many object-oriented methodsin a more mature form than IIDEF4, with CASE tools to support applications, unless the advantage gained through its links with other IDDEFmethods is significant widespread adoption of the method seemsunlikely. 252 APPENDIX4 MODELLING METHODS REVIEW 253 CONTENTS A4.1 THE CHECKLAND SOFT SYSTEMS METHODOLOGY 254 ...... A4.2 THE VIABLE SYSTEMS MODEL (VSM) 256 ...... A4.3 THE GRAI METHODOLOGY 257 ...... A4.4 TBE STRUCTURED SYSTEMS ANALYSIS AND DESIGN METHODOLOGY (SSADM) 259 ...... A4.5 TBE NUSSEN INFORMATION ANALYSIS METHOD (NIAM) 260 ..... A4.6 TBE CONTROLLED REQUIREMENTS EXPRESSION (CORE) METHOD 261 ...... A4.7 DISCRETE EVENT SIMULATION 264 ...... A4.8 PETRINETS 265 ...... A4.9 CRITICAL PATH ANALYSIS (CPA) 266 ...... A4.1 THE CRECKLAND SOFT SYSTEMS METHODOLOGY At the clearly'soft' end of the modellingspectrum shown in figure 4.1, the Checklandsoft systemsmethodology (Checkland, 1981, Wilson, 1984, Checklandand Scholes,1990) is proposedas a meansof understandingsituations by developinginsight into how participants in interact the 'From a systemview and with system. a soft systemsperspective , the collectiveperceptions of the membersof a systemare the systemfor the peoplein that system'(Cavaleri and Obloj, 1993). The methodologyprovides an analystwith a meansof structuring an investigationof systemsproblems. In brief the complete methodology comprisesseven stages: 1. Recognitionof the problemstation. 2. Building a 'Rich Picture' of the situation. 3. Establishinga 'Root definition' to provide a precise text based descriptionof a situation that embodiesthe elementsof the problem describedby the mnemonicCATWOE where; eC describesthe receiversof the resultsof the transformation. 254 "A describesthe 'actors' who carry out the transformation. "T the transformationprocess or activity that embodiesthe problem situation. "W is the viewpoint or 'worldview' implied in description 0 describesthe owner who controls the transformation. E describesthe environmentor natural constraintsof the system. 4. Fromthe root definitiona conceptualmodel of a systemcan be producedin graphicaland text form to providean ideal 'TO BE' descriptionof the situation. 5. A comparisonof the idealised'TO BE' modelwith the 'Rich Picture' of the existingsituation. 6. Analysisand discussionof the issuesraised by comparison,with a view to proposingchanges. 7. Action to implementchanges. The relativelyunstructured diagrams resulting from applicationof the methodologycannot easilybe comparedwith the rigorous,formal presentationof other modellingmethods such as GRAI, Petri nets or 11DEFmethods. At a high level of abstractionhowever an 113EFO modelrepresents a systemin a similarway usingthe 'Context' and 'Viewpoint' diagramsfor AS IS and TO BE modelsand wherethe investigationto establishICOMs can be loosely comparedto a CATWOEroot definition. CVstomerscan be seenas the recipientsof IDEFo outputs. Transformationsas IDEFOfunctions. Actors and Ownersas IDEFomechanisms and controls. InterestinglyWu (1994,p 210) proposesIDEFo as a modellingmethod that could be usedat stagefour in the ChecklandSoft Systemsmethodology as a diagramming methodfor a conceptualmodel. In reviewing developmentof the methodology since its inception (Checkland and Scholes, 1990) reflect that 'SSM not only developsand changesbut also gets used in different ways by different users in different circumstances.' and the approach although not specifically developed for manufacturing enterprise applications has been used to solve manufacturing problemsCheckland (1989). 255 A4.2 THE VUBLE SYSTEMS MODEL (VSM) sysbma PO/ICY Marwgkig DVactor maft SysftM 4 Seff rafar&vo, ahwAsdon, Group plarming,R&D pwvft SYstem 3 cor&d AUDffCHAAWEL COMROL kftmafdata Rume, Saiss gagwed for 4W&m 5 LOCAL ENWROWENT "-ýýCo-ordftuMw 0 Syxbm 2 &4ý Monhodng and AY,W leguAmbon to prevent system I OWIIABUVout Of LU r1 control Non-EngtneMng Mordtodng SYsbm I ft opamdng cam Yft /L*J.,-d M-agemem Figure A4.1 The Viable SystemsModel (adaptedfrom Espejo, 1989b) The Viable SystemsModel (VSM) approach developedby Beer (1979,1981,1984) has evolved from OperationsResearch (OR) and Cybernetics. Beer perceivesthe human nervous systemas his exemplar,the nervous systemfollows a number of key principles to 256 ensureits survival,the brain has to regulatea large numberof variables(temperature, balanceetc. ). It has to learn to adapt to changingsituations and develop over time. Analogiescan be drawnwith the control andmanagement necessary for largeorganisations suchas manufacturingenterprises. I-Es approach is basedon the preceptthat there'arefive generic sub-systemsinteractively involved in any viable organisationinvolving human beings;operating core, control and co-ordinatingmechanism, internal information systems, externalinformation systems and policy makingthey are shownin an examplein figure A4.1. FigureA4.1 hasbeen adapted to illustratethe VSM in a manufacturingdomain. The VSM has been developedas a general systemsapproach to develop strategiesor organisationalpolicies and manufacturingapplications have been cited by Espejo and Hamden(1989) and Kehoe et al., (1993). A4.3 THE GRAI (Graphe i Risultats et Activit6s Interhis) METHODOLOGY (Graphs with results and activities interrelated) The GRAI methodology (Doumeingts, 1984, Doumeingts, 1985, and Pun et al, 1985) has been developedto 'analyseand design CIM systemswith particular emphasison Production Managementand Flexible Manufacturing Systems' (GRAI Laboratory, 1987). The GRAI approachin commonwith IDEFo and unlike many modelling methodshas been developedto deal specificallywith manufacturing. The GRAI conceptualview of manufacturingorganisation proposes that there are three sub- systems,the 'Physical system' consisting of the means of production, men, machinesand material converting raw material into products. Above the physical system is a 'Decision system' a hierarchy of decision centres, at the top of the hierarchy are long term strategic decisions and at the lower levels of the hierarchy are the real time decisions that directly control manufacturing. Finally, connectingthe Physicalsystem and the levels of the Decision systemhierarchy, is the 'Information system'. Application of the method produces a graphical model that representsthe operation of each decision centre in terms of its activities, the time frame of operation, the decisionsmade and 257 the information used. Graphical representation is based on two types of diagram, GRAI grids and GRAI nets. The GRAI grid is, a top down descriptionof the structureof decisioncentres showing a hierarchyof time horizonsand periodsas rows in the grid and functionsas columnsin the grid (seefigure A4.2). INFORMATION TO TO TO MANAGE PLJRC+P.SE PLAN TECMCAL INTERUAL RESOLRCES I-F SCTERNAL Hw1Y Orders Budget MPS N3M z --100, j - WM Purchase Scheduling N3d material& comporots Hx1w Shopflw scheduling N3d REAL 'nME Executing plan Figure A4.2 A GRAI grid for a 'make to order' company. Each block of the grid representa potential decisioncentre, flows of infonnation are identifiedwith singlearrows and the transmissionof decisionsshown as thick arrows. A GRAI net (an exampleis shown in figure A4.3) is a form of decompositiondiagram that relates decisionsidentified in the GRAI grid to the information and resourcesrequired to executethe decision. For a detailed account of the application of Nets and Grids the reader is referred to 'The rules of the method and analysis of GRAI Grids' (GRAI Laboratory, 1991). 258 Decomposition to expose greater detail is accomplishedby 'exploding' an overall (global) grid into detail grids. In comparing GRAI and IDEFo some fundamental differences are evident. The GRAI method is focused on the identification and analysisof decision centresand in the GRAI grid only 'important' information is identified. In contrast to the central IDEFo focus on activities andthe relationshipof activitiesin termsof informationor objects. Figure A4.3 A GRAI Net (source:Roboam and Pun, 1989) A timescale in the form of planning horizons and planning periods (in effect the planning horizon for a decision and the length of time a decision is effective) is a dimension in the GRAI grid that cannot be explicitly representedgraphically using IDEFo. A4.4 1711ESTRUCTURED SYSTEMS ANALYSIS AND DESIGN METHODOLOGY (SSADM) In the LJ& SSADM is a widely accepted,structured methodology (Cutts, 1991, Longworth and Nicholls 1987) that utilises severalgraphical modelsfor the analysisand designof 259 software systems. The complete methodology is based on three phases;feasibility study, Systemsanalysis and Systemsdesign and comprises eight stages;problem definition, project identification, analysis,specification of requirements,selection of systemoptions, logical data design,logical processdesign, physical design. The methodology uses graphical modelling in the form of data flow diagrams and entity- ýelationshipand entity life history diagrams. The data flow diagramused in SSADM is based on the older DFD technique propounded by DeMarco (1979). In SSADM the data flow diagramrepresents activities or functions using rectangles,the sourcesor destinationsof data using ellipses,data flows using arrows and data stores using open-endedrectangles (shown in figure 4.6). The detailedrules for model building advisethat there should be no more than three levels of decompositionin contrast to IDEFo's unlimited approach. EDEFocannot be directly comparedwith the complete SSADM methodology however the SSADM data flow diagrmn has its roots in the DFD described by DeMarco and as such comparisonof the basic principles of data flow diagramsand IIDEFois valid. A4.5 THE NUSSEN INFORMATION ANALYSIS METHOD (NIAM) The NLAM informationmodelling method has been developedin Europe and Australia (Nijssenand Halpin, 1989)and is widely usedin ESPRIT projects(Los et al, 1992). The methodis orientedtowards situations where an informationprocessing system is used to communicatebetween users and an object systemsuch as a manufacturingsystem. A completereview of the diagramsused in the methodand the NIAM methodologyis givenby Nijssenand Halpin (1989) which includes all the stepsnecessary to get from a NIAM model to a relationaldatabase. There are someimportant Merences between NIAM andthe more establishedEntity/Relationship methods such as IDEFix- A fundamentalconcept is that an Information StructureDiagram (the NIAM equivalentof an E/R, diagram)distinguishes betweenNon-Lexical Object Types (NOLOTs) which are real life objects and their descriptivedata - Lexical Object Types (LOTs). NOLOTs and LOTs are equivalentto 260 entities in conventionalE/R modelling but NLAM doesnot attach attributes and consequently doesnot apply nonnalisationto databasedesign. A4.6 THE CONTROLLED REQUIREMENTS EXPRESSION (CORE) METHOD The design of CORE (Curwen, 1990) was funded by British Aerospace as a method to develop systemand software requirementsspecifically for Aerospacemanufacturing. Some aspects of the method are based on SADT and Parnaby et al (1988) claim that CORE provides 'some useful additions to IODEFo. Application of the CORE method involves producing seventypes of diagrams: " Viewpoint structure diagrams " Tabular entry diagrams " Data composition diagrams " Isolated thread diagrams " Combinedthread diagrams " Isolated operationaldiagrams 40 Combinedoperational diagrams The diagramstogether with structured text descriptionscalled 'Node Notes' supplementing each diagram form a complete CORE model. The diagramsof particular relevanceto this researchare the Viewpointstructure diagrams and the variousthread diagrams. Viewpoint structure diagrams are the result of a 'viewpoint analysis' to establish all the viewpoints to be consideredin the problem. The viewpoints are representedin a hierarchical tree developed using a set of rules based on partitioning viewpoints into user views, functional views, non-functionalviews, data views etc.. Tabular entry diagrams are used to collect information on the processesinvolved in each viewpoint. Figure 4.4 is an example of a Tabular entry diagram showing structure of the information coHectedfor eachprocess (a processis tenned an 'action' in CORE). 261 Figure A4.4 shows one (prove program) of the many actions that could be involved in the Bridgeport milling machine viewpoint and the four categories of information necessaryto completethe tabular entry for the action. Tabular collection diagramsdescribe input and output specificationsand identify the actions in a particular viewpoint ( the viewpoint in figure A4.4 is that of the 'Bridgeport milling machine'). For every action in eachviewpoint an Isolated thread diagramis then produced. Figure A4.5 shows a thread diagram for the 'prove program' action in figure A4.4. Thread diagramsare similar to DFI)s and IIDEFodiagrams. Arrows entering the left side of boxes areinputs, outputs are on the left side,boxes describe a sequenceof actionsand the sources of diagraminput arrows and destinationsof diagram output arrows are labeHed. BPJDGEPORTMILLING MACHINE VIEVVTOINT SOURCE INPUT ACTION OUT? UT DESTINATION I,,,, It _hfill Planner Prove NC Program -Reject - -Planner program ý'Clash Planner warning Figure A4.4 A CORE tabular entry diagram 262 Planner ProgramOK I- I NC program -1 Am9Pt Milling Load nin tool path Download Mcn. progi simulation to Mcn. Irwald program 0- Reject Request------11 Planner program CJash change warning Figure A4.5 An individual thread diagramfor the 'Prove program' action Figure A4.5 illustrates someof the graphical conventionsof CORE : " The ordering of boxes from left to right is significant and indicatesthe sequenceof - actions. " Control information is shown entering or leaving the top of boxes. " If boxes are alignedvertically actions may be carried out concurrently (not in the caseof figure A4.5). No partial alignmentoverlap of actionsis allowed. "A dotted line indicatesa control that must be triggered by an event, in this caseit could be demandfor a machinedcomponent. " 'Critical'data is shown with a dotted line, where critical data is 'eventually used and used only once' and 'consumedwhen read and not overwritten'. " 'Non-critical' data is shown with a solid line, is data that 'overwrites previous data' and 'can be read many times before being refreshed'. In figure A4.6 someimportant featuresof action box control syntax are illustrated. 263 Cor*W dab canWso contain or wkm Iriftmation usedto seled HX&ALYmiý actions Actfonbox nLmnber 0 AcWm ~ Action C omw comnuously ActionA dbpAsyan &Wadsk how mfinb rwmber of m exclusion4mW N kwaaws V* W'SnnokW f---- numbw Mmnber of an Vw flinib ftmoan symboL Fm*o Awation symbol Aotwx whth wo mLtme 13 OWJUMO(&«wffi4 am vwtiu* aNnad wm 10- r, whw oa shoffl Action D An adon In continuous urd to nka of'DATA-0- CNmges asw UNTIL-FALSE D Figure A4.6 Examplesof CORE action box syntax (Adaptedfrom Curwen,1990) The symbol 'D' used in the top centre of an action box indicatesthat an action box has been decomposed,decomposition involves the sameparent/cud syntax checking as IDEFo. 'Mechanism' arrows are shown entering the lower edge of an action box. In CORE terminology a 'mechanism' can be the viewpoint name to which the action belongs or an implementationpartition. Combinedthread diagrams are a compositeof Individualthread diagrams used to describe the relationshipsbetween the different viewpoints being considered. A4.7 DISCRETE EVENT SIMULATION Computerbased discrete event simulation uses mathematical models to predictthe changesin a systemas it evolvesover time. 'Events' are points in time when an instantaneousoccurrence may changethe state of the system. The input to models is stochastic,based on statistical 264 distributions and will produce an output that is in itself random and is therefore an estimateof the behaviour of a model. Simulation could be done using manual calculationswithout the use of formal graphical descriptionsbut the complexity of manufacturingsystems and the amount of data to be stored and manipulateddemands the use of a computer. The developmentof computer graphics,RAM and hard disk memory capacity over the last ten years has led to the routine use of animateddisplays for events and the states of models. The graphical system description aspects of discrete event simulation are, in many cases, incidental to the mathematicalmodels underlying modelling, although they are now an essentialuser interface for system definition and the interpretation of modelling results. For this research discrete event simulation was considered outside the context of static graphical manufacturing enterprisesystems modelling. A4.8 PETRI NETS Petri nets (Petri, 1980) have been developed to model systems such as cornmunication networks and information systems and have the ability to represent serial and concurrent events(Peterson, 1981). The Petri net method has been applied in a wide range of fields and has been used extensively to model manufacturing systems (Eswara Reddy, et al., 1992, Zhang, 1989,DiCesare et al, 1993). A Petri net diagramis a directed multi-graph representing events,the conditions that control events and the relationship between events and conditions. The method is mathematical and is 'specially suited to model and analyse discrete event dynamic systems'(Silva, 1993) but unlike simulationthe graphical representationis an integral aspect of the model. In the context of manufacturing systemsa diagram consists of circles representingthe conditions of a system (places), bars representing events (transitions), the state of a condition is representedby a dot (a token) in the specified'place' circle and directed arcs (arrows) show the direction of material or information flow between 'places' and 'transitions'. Much work has taken place to develop a body of research high level predicate/transition nets (Genrich and Lautenbach, 1981) and subsequentlyColoured Petri Nets (Jensen, 1990). The fundamentalPetri net method characterisedby Peterson (1981) does not attempt to model 'time' although researchwork has been carried out to extend the 265 initial work using a time dimension for transitions or places (Son, et al. 1991). For the purposesof this analysisthe basic graphical representationis considered. A4.9 CRITICAL PATH ANALYSIS (CPA) CPA (Moder and Phillips, 1970) is a modelling method to solve sequencingproblems based on identifying a logical sequenceof activities or tasks and identifying the events (beginning and ends)associated with those activities together with the time betweenevents. It is essentiallya mathematical approach but the graphics used to describe events and relationships are an integral part of the method. It is used in three basic forms, activity on arrow, activity on node and the ProgrammeEvaluation and Review Technique(PERT), for the purposesof the review in chapter four the activity on arrow approachis considered. Although CPA is not normally consideredin a discussionof manufacturingenterprise modelling methods CPA is a graphical structured analysis and design method capable of modelling the logic of events in such systems. In contrast to the more recently developed systems modelling methods CPA is universally accepted,established and routinely used in a wide range of applications including manufacturing systems. It is normally considereda Project Managementtool for large, 'one of a kind' civil and mechanicalengineering projects. However in an interesting application in the shipbuildingindustry Wade and Karaszewski,(1992) have proposedan approachthat uses CPA to analysefunctions in an 11DEFomodel. Its strength in terms of a comparative analysis lies in its accepted successand routine use as a graphical modelling method, its usability, accessibility, descriptive capabilities and its potential as a benchmark for the acceptanceof manufacturingenterprise modelling methods. 266 APPENDIX 5 EPSRC-CDP GRANT ANNOUNCEMENT 267 Engineering and Physical Sciences EPSRC Research Council RESEARCHOFFICE Polaris House DERBY UNIVERSITY North Star Avenue KEDLESTONROAD Swindon SN2 1ET DERBY Telephone 0793 444000 Central Fax 0793 444010 DE22 1GB Tel (Direct Line): Grant Ref: GR/K39400 Date: 02 September STANDARD RESEARCH GRANT ANNOUNCEMENT In respect of Grant Condition Itt (b) Commercial Exploitation through the BTQv applies Dear Sir/Madam The Council is prepared to make a Standard Research Grant towards the cost of the research proposed by the Investigator(s) named below. Payment of the Research Grant will be made subject to the Council's Research Grant Conditions listed below and any additional conditions specified. The Council must be informed within one month if the research grant is not acceptable to your Institution on these terms. The Institution must confirm acceptance of this research grant and inform the Council of the actual start date by return of form SD1 within 28 days of the actual start date. Signed SECRETARIAT, ACME IMPORTANT: Please pass the enclosed copies to the Investigator(s) named below. (Off ýe4oakd-f GRANT PERIOD: 36 Months STARTS: 01/10/94 ENDS: 30/09/97 INSTITUTION: 6070 DERBY UNIVERSITY PRINCIPAL INVESTIGATOR: PROF RW BAINES SCHOOLOF ENGINEERING PROJECT TITLE: BEST PRACTICE MODELLING METHODSIN MANUFACTURINGENTERPRISES L Months Posts Payment limits STAFF RA Range 1B 50,754 36.00 1 FY L RA Range 1A 0 0.00 0 94/95 8,386 Technician 0 0.00 0 95/96 33,544 Other 10,500 12.00 1 STAFF TOTAL 61,254 48.001 2 TRAVEL & SUBSISTENCE 19,814 CONSUMABLES 2,150 Further details on EXCEPTIONAL ITEMS 8,400 staff posts awai ded EQUIPMENT 3,756 appear below SUB-TOTAL 95,374 INDIRECT COSTS 24,502 Rate of indireci GRANT TOTAL 119,876 costs = 40% Page 1 of EPSRC RESEARCHGRANT ANNOUNCEMENT GR/K39400 02 September 1994 CO-INVESTIGATORS MR GJ COLQUHOUN LIVERPOOL JOHN MOORESUNIV ENGINEERING & TECHNOLOGYMANAGEMENT STAFF POSTS Spine Grade & Date of Start Increment Grade Pt Scale Posts Months Date Date Name / Comments 3 RAlB 2 01/04/93 1 36.00 01/10/94 01/10/95 OTHER 1 12.00 01/10/94 SECRETARIAL Page 2 of 4 269 APPENDIX 6 KBSI RESEARCH GRANT AGREEMENT 270 KNOWLEDGEBASED SYSTEMS, INC. February3,1994 Mr. Gary Colquhoun Liverpool JohnMoores University Fax 44-51-298-2447 Dear Mr. Colquhoun: Ms. Caley McBroom is out of town attendinga conference,so I will be handling your ordering procedure. I am pleased to inform you that conditions for the Acaden-dc Grant have been met and your Acaden-dcpackage will be sent today by International Federal Express. I only ask that you pIck up the shipping and duties. A bankers draft will be acceptable and may be made out to the amount listed below In U. S. dollars. The total cost of the package is as follows: PROSIM $5995.00 Grant -$5840. TOTAL $ 155.00 TechnicalReports $ 99.95 ShippingtHandlingof FedExInternational 7 IN Package $ 92.00 GRAND TOTAL $ 346.95 (three hundred forty-six dollars and ninety-five cents) An Invoice will, be attachedfor your convenienceIn mailing. As always,customer satisfaction is our priority. Pleaselet me know If you have any questions. Sincerely, Sally Penick Marketing Representative OneMISI Place 1408University Drive East CollegeStation, Texas 77840-2335 409.260.5274 Fax:409.260.1965 271 APPENDIX 7 IDEF3 MODEL OF'RUN ZENO' 272 %0 Z %0 94 mm e4 JL4 Figure AM IDEF3 diagram 'Run ZENO' 273 Z! "p I q4 Figure A7.2 IDEF3 diagram'Inspect and pack product' 274 r! LN 5 - -ST 3 44 OX I I E jII,,,(I II;I 1;: 1, 1tI- :,.E" III 1 21 if - .2' , 2-9 i ei92 Figure A7.3 IDEF3 diagram'Monitor process' 275 Pt! e4 x sc Figure A7.4 IDEF3 diagramRun 400 press' 276 %0 Figure A7.5 IDEF3 diagram 'Set up 400 press' 277 Na N Figure A7.6 IDEF3 diagram 'Carry out batch run' 278 APPENDIX8 THE MODELLING METHODS QUESTIONNAIRE 279 I g PIIEImI . I tt Z) c m5 -x-- u _C) c» c -U 11 11 1 4 . - - 0 u vi L.. 0 C _C 0 0 & cc Q) > 0 Cli 117 c: c C', S c! 0 El 0) C vi M r-4 C) (L) (Z I Lj c F E I , e E. 0 0 E A: -,ý = Eiý - I O m. 5; v V > ni 'C' j :3ý61 I v! 0 0> c ol m El -C 1.1 v c 1 C, 2 FU ab Cý e-h -0- = 0 %0 cEC: (u C, Cl CD 0 VWMM0 C0 vý Q '0EL2 0 LoNf5p, 01SC rýIc P--.u ä - ru- u I w.c: ) r_ Q) -,% c -0, Wl 04 0 ý E c ig i S C. 1: _C):3 Z0 01, * 06I wL u. LI) ci uj Ln 01 c: LL)Lu 0 .c OZS t t, tm L cs c 2) 1 j C c LVI ý m c JD Z E E 10 l m 1ý tu 0 - -Z ýo E Ela c v Y0 :3 'D u c 0 0) LAL2- t c .r- 7 1 1 t E -C I :) a I Li VI G? q) E o týL E w i -M ý c . V) 21 C C LA 0 UE It V S 4) E C: 5; -r- ri c LZ11)iLý: j r- iuC, C, :e < c ý, zQJ <'5 c (ý,0 O 0 4 o i LZ u iY- lco -5 ýý et .W Figure A8.1 The modelling methodsquestionnaire 280 APPENDIX9 PROCESS PLANNING FOR BATCH NUNUFACTURING 281 PROCESS PLANNING FOR BATCH NUNUFACTURING In a batch manufacturingpiece part production environmentthe processplanning activity plans the manufacture of components and products and involves the design or re- configuration of manufacturing methods, establishing the necessary resources and capacity (tooling, equipment, manufacturing times, staff, material, etc.). A process planner receives component design information detailing component geometry and attributes such as surface finish, tolerance hardness,etc., together with batch sizes. He applieshis experienceand usesavailable data to produce a processplan (the actual terms used are often company specific, i. e. operation layouts, route sheets,planning summary, etc.). The process plan provides a sequence of manufacturing processes,operation details for eachprocess, set up and cycle times and tooling, fixture and gaugedefinition. In the UK generally the decline of manufacturinghas meant changesin organisation and staff reductions and a shrinking population of skilled staff. Traditionally processplanning was seen as a separate,de-coupled function that took place after product design. The decision making processesinvolved in deriving a processplan are complex, the planner relies on his manufacturing experience and accessto standards and previous process plans. To make manufacturingdecisions, the shape,size, range and finish that a process is capableof producing must be known. To identify a specific machine,the power and capacity must be balancedwith raw material (specification and size) and tool properties (wear rates, geometry, etc.). Reducing skills on the factory floor, has resulted in a need to provide more comprehensivemanufacturing instructions by fewer planners for ever more technologically complex manufacturing. The result can be process plans that are inconsistent,inaccurate and out of date due to engineeringchanges, batch size variations and changesin availabletechnology. A recognised.problem of process plan design is the difficulty in maintaining consistent plansfor the samemanufacturing features when createdby differentplanners and, to a lesserextent, plans created by the sameindividual. Studieshave been carried out that show the diversityof decisionmaking encountered with the subsequentproliferation of 282 tooling, fixtures machines and process routes. Process planning has a high clerical content both searchingfor data and document preparation with as little as 20% of the plannerstime being spent in technical decision making. It is in the clerical activity that computer assistanceto the planner can provide the most immediate, improvement using word processingfacilities, standardplan formats and a common engineeringdatabase. The processplanning activity itself has also changedas Computer Aided ProcessPlanning (CAPP) systemshave developedfrom simple databasesthrough retrieval systems(Chang and Wysk, 1985) to the semi-generativesystems (Zhang and Alting, 1994) commercially available today. In a computer based integrated manufacturing environment process planning is an essential element in the flow of information between Computer Aided Design (CAD) and Computer Aided Manufacturing (CAM). There is recognition that CAPP has the potential to provide the essentiallink between CAD and CAM (Alting 1986, Chu and Wang 1988, Bok and Nee 1988, Sutton 1989, Wang and Wysk 1988). The aim of CAPP system developersis that systemswill be able to take the geometric description of a component, apply planning logic and then produce the manufacturing instructions without humanintervention. Much researchand developmentwork has been carried out in the area, in 1989 Alting and Zhang claimedthat more than 200 papershave been published on the subject since 1970. Commercial developmenthas led to CAPP systemsbecoming available and being used by major manufacturersin the UK. 283 APPENDIXIO PART NUMBER/PROCESS MATRIX 284 S ý - w '010 M I r: T P m If - ------ R181 5ý IR 1 1 2 R I I C. I 9 8 s 10 8 S R-AR M 8 9 9 R PIR S S M 8 - s $ 8 8 -2 A Pl L I, < < Qm < ;E -2 I a: - - " 44 " " - - < - - " " ------1 Q, z it i 9 § ý I a R Po &, z 44 M V ý$ gi. ý"K IN I 'l , i -1 a I1 IAN l " I I I I-71 I ,i i i Awý - -l -Iý-W 1- Table A10.1 Part number/Processmatrix APPENDIX 11 BSD SCHEDULING/AL4, NAGEMENT SOFTWARE SPECIFICATION 286 SOFTWARE SPECIFICATION SCHEDULING/MANAGEMENT BSD STEEL SERVICE CENTRES ...... 11INTERNAL USE Oý. Lyj Toronto Industrial Fabrications Dunlop & Ranken Estate Whitehall Road Leeds LS12 6JG Tel: (0113)2311911 Fayý.(0113) 2311982 Shect I of 4 ...... The above shouldbe completewith the following as minimum for stagesI&2 of implementation- Control will comprise of three basic functions: Produce schedules, documents and data. Control work centre schedules. Track inventory. Operatingplatform, environmentand users. PC based. Windows environment. 4 user systempassword protected. 2 read only (a) - with Vhat if scheduling capabilities, i. e., to be able to examine and experiment with schedule, without the authority to issue or changethe masterschedule. I read only (b) - with ability to produce picking document,no 'what ifs'. 1 full user - with authority to examine, issuelproduce schedules and documents. SYSTEM REQUIREMENTS 1. Produce Scheduleand documents(format to be designedby BSD with A4 size documents). 1.1 Work centre schedules- Daily for eachwork centre. 1.1.2 Work centre number. 1.1.3 List of part numbersin sequenceof loading. 1.1.4 Batch/works numberagainst each part. 1.1.5 Date of issue. 1.2 Route card - Per batch. 1.2.1 Part numberand description. 1.2.2 Material specificationand thickness. 1.2.3 Graphicaloutline of part via DXF import. 1.2.4 List of work centresin order of processingand alternateroute alongside. 1.2.4.1 Batch/works number. 1.2.4.2 Quantity in batch. 1.2.4.3 Date of issue. 1.2.4.4 Description and numberof work centre. 1.2.4.5 Cycle times at work centre. SAH. 26104195 1.2.4.6 Sign off column at eachprocess with scrapand rework noting. 1.2.4.7 Note column (e.g., tooling). 1.2.4.8 Acceptednest number. 1.3 Sub-contractschedules and purchaseorder. 1.3.1 Quantity in batch. 1.3.2 Date of issue. 1.3.3 Batch/works number. 1.3.4 Part numberand description. 1.3.5 Due date, to be returned to BSD works. 1.3.6 Signaturesfor dispatchand return. 1.4 Dispatch/pickingdocument (3 Ply). 1.4.1 Purchaseorder number. 1.4.2 Kit number. 1.4.3 List of componentsin Kit with quantitiesand batch/worksnumber. 1.4.4 Signatureacceptance for driver and customer. 2. Managementinformation. 2.2 Resourceutilisation (work centresand staff). 2.3 Routings, alternativeroutes and Bill of materials. 2.4 Work in Progress(KY), shop status. 2.5 Overdue!s and delays. 2.6 Costs. (May be displayedvia 3rd party database) 2.6.1 By resource. 2.6.2 By works order number. 2.6.3 97P and Inventory on shop floor (and Projected). 3. Rough cut schedulingfor profiling only - Over 12 week horizon. 3.1 Data file output to nestingsoftware. 3.1.1 Part number. 3.1.2 Due date. 3.1.3 Quantity. 3.1.4 Material grade. 3.1.5 Thickness. 3.1.6 Plasmaor gas operation. horizon. 4. Detail schedulingfor profiling only - Daily with up to 2 week 4.1 Data file output to nestingsoftware. 4.1.1 Part number. 4.1.2 Due date. 4.1.3 Quantity. 4.1.4 Material grade. 4.1.5 Thickness. 4.1.6 Plasmaor gas operation. 4.1.7 Hi/Lo priority. S. Control work centre schedules. 5.1 Demand/customerschedule for kits (level I items). 5.1.1 Quantity of kit. 5.1.2 Due dates. 5.2 Produce due datesand quantities(work to list) for individual part numberswithin kit ordered (level 3 items). Sheet 2 of 4 SAH. 26104195 298 5.2.1 Via BOM explosion and part routings. 5.3 Produceinfinite capacitywork centreutilisation using backward scheduling. 5.4 Allow use of alternateroutes/resources via manualinteraction. 5.5 To be capableof running in 2 modes. 5.5.1 Rough cut scheduleto plan gross plate requirement for profiling only - 12 week horizon, not to include 'on-hand'inventory. 5.5.2 Detail schedulingto plan due date and quantities for all work centres, including daily basis. profiling -2 week horizon on a re-plan 5.5.2.1 System must have ability for user to adjust capacity, resources (alternate routes), quantities and due dates in order to balance customer demand to availablecapacity. This should be on a 'what-if interactive graphical display type applicationwithout compromisingmaster schedules. 6. Track inventory. 6.1 Calculationof IM. 6.2 Finishedcomponents. 6.3 Scrap& Rework. 6.3.1 Scheduleand inventory level must be updated. 6.4 Over production quantities (with part numbers) from nesting with relevant schedule adjustment. 6.5 11istoricalproduction. 6.5.1 Quantitiesmanufactured to date. 6.5.2 Capacitiesand utilisation's(resources) to date. 6.5.3 Late orders to date (inc. remainingoutstanding jobs). 6.5.4 Dispatchedorders. 7. Inputs to system. 7.1 Dynamic inputs. 7.1.1 CustomerKit schedules(kit requiredwith due date on a roll out diary). 7.1.2 Exceptionson resources. 7.1.3 Alterations on inventory level. 7.1.4 Scrapand rework -,krith reason codes and work centre. 7.1.5 Feedbackfrom nestingsystem via data file, after detail scheduleonly. 7.1.5.1 Part identification againstquantity nested. 7.1.6 Alterations to plannedschedule. 7.1.6.1 Delays on batches,system to assumeorder completed when given time has elapsedunless otherwise stated (Stages I&2 only). 7.1.6.2 Uncompletedorders. 7.2 Fundamentalinputs. 7.2.1 Resourcecapacities (machines and personnel)with calendarhours. 7.2.2 Fixed costs. 7.2.3 BOM. 7.2.4 Processroutes and alternateroutes, inc. sub contract possibilities. 7.2.5 Part groupings(for like tooling). 7.2.6 Part Classifications. 7.2.7 Kanbanre-order levels(inventory file links). 8. Systemconfiguration and backup. 8,1 Systemto be tailored to BSD requirementsover 3 month implementationprogramme, with start to be negotiatedafter order placement.' 8.2 Subsequentchanges to be madepossible with BSD to cover costs. Sheet 3 of 4 S.R. H. 26104195 289 8.3 Operator training, for up to 4 personsif required. 8.3.1 To be split, with training for I initial user, and follow up for additionalusers. 8.3.2 Follow up internal training to be by BSD personnel. 8.4 Maintenancecontract. 8.4.1 Modem link, with. security access. 8.4.2 Help desk. 8.4.3 Any re-training due to changesin systemnot specifiedby BSD. 8.5 Manuals. 8.5.1 As systemis to be configured, manualsto be written by BSD personnelin terms of useability(visual), proceduresand interpretationof results. 8.6 Consultancy. 8.6.1 To be readily availableduring installationphase, with ongoing requirementsat a cost to BSD. STAGE 3 DEVELOPMENT 1. Data collection from shop floor (real time). 1.1 Shop floor terminals. 1.1.1 Bar-code labels. 1.1.2 Bar-code scanningequipment. 1.1.3 Hand held data collection. 1.1.4 Dumb terminals. 1.2 Shop floor documentprinting. 1.2.1 Reports. 1.2.2 Bar-code labels. 1.3 Control of scrapand rework in real time. 1.3.1 Bar code sheetswith reasoncodes for error/problem. 1.4 Productive measurementsof personnel. 1.4.1 Bar code sheetswith reasoncodes for non-productivetime. DIAGRAMS ATTACHED 1. CENIEEscheduling (Requirement schematic). 2. Information Models. 2.1 A22 - Establishmanufacturing schedules. 2.2 F80 - Establishmanufacturing schedules. 2.3 A223 - Control work centre schedules. 2.4 A221 - Establishnesting data. Sheet 4 of 4 S.R. H. 26104193 0c 0 r_ rc u 4, 0 y) 0 (L) mu 0) C) L) UU U) (1) L) 6D -5 cl- 5; 32 0o .0u - ad L4 0< 10 >m 0Z IIPf4IP*I 0 ". U) ad C) :)00 c C od 4) 0U U) L4 :8, 0m lz c CL Sý !; 3: C 0 CD tu U u U -0 m U, Z, Q) u ýL. :x C:L, A. mmmýrz N Z go CZ r1) ..CA C? '. 44 Cfi cl A 7 C) 00 V-4 - -E 00 00 -) iz. Q) _r_ ýz 291 PAGE/PAGES EXCLUDED UNDER INSTRUCTION FROM UNIVERSITY