Discrete-Event Simulation and System Dynamics for Management Decision Making

Sally Brailsford, School of Management, University of Southampton, UK Leonid Churilov, Melbourne Brain Centre, Victoria, Australia Brian Dangerfield, Salford Business School, University of Salford, UK

• Preface xvList of contributors xvii1 Introduction 1Sally Brailsford, Leonid Churilov and Brian Dangerfield 1.1 How this book came about 11.2 The editors 21.3 Navigating the book 3References 92 Discrete-event simulation: A primer 10Stewart Robinson2.1 Introduction 102.2 An example of a discrete-event simulation: Modelling a hospital theatres process 112.3 The technical perspective: How DES works 122.3.1 Time handling in DES 142.3.2 Random sampling in DES 152.4 The philosophical perspective: The DES worldview 212.5 Software for DES 232.6 Conclusion 24References 243 Systems thinking and system dynamics: A primer 26Brian Dangerfield3.1 Introduction 263.2 Systems thinking 283.2.1 ‘Behaviour over time’ graphs 283.2.2 Archetypes 293.2.3 Principles of influence (or causal loop) diagrams 303.2.4 From diagrams to behaviour 323.3 System dynamics 343.3.1 Principles of stock–flow diagramming 343.3.2 Model purpose and model conceptualisation 353.3.3 Adding auxiliaries, parameters and information links to the spinal stock–flow structure 363.3.4 Equation writing and dimensional checking 373.4 Some further important issues in SD modelling 403.4.1 Use of soft variables 403.4.2 Co-flows 423.4.3 Delays and smoothing functions 433.4.4 Model validation 463.4.5 Optimisation of SD models 483.4.6 The role of data in SD models 493.5 Further reading 49References 504 Combining problem structuring methods with simulation: The philosophical and practical challenges 52Kathy Kotiadis and John Mingers4.1 Introduction 524.2 What are problem structuring methods? 534.3 Multiparadigm multimethodology in management science 544.3.1 Paradigm incommensurability 554.3.2 Cultural difficulties 574.3.3 Cognitive difficulties 584.3.4 Practical problems 594.4 Relevant projects and case studies 604.5 The case study: Evaluating intermediate care 624.5.1 The problem situation 624.5.2 Soft systems methodology 644.5.3 Discrete-event simulation modelling 664.5.4 Multimethodology 674.6 Discussion 684.6.1 The multiparadigm multimethodology position and strategy 684.6.2 The cultural difficulties 704.6.3 The cognitive difficulties 704.7 Conclusions 72Acknowledgements 72References 725 Philosophical positioning of discrete-event simulation and system dynamics as management science tools for process systems: A critical realist perspective 76Kristian Rotaru, Leonid Churilov and Andrew Flitman5.1 Introduction 765.2 Ontological and epistemological assumptions of CR 805.2.1 The stratified CR ontology 805.2.2 The abductive mode of reasoning 815.3 Process system modelling with SD and DES through the prism of CR scientific positioning 825.3.1 Lifecycle perspective on SD and DES methods 845.4 Process system modelling with SD and DES: Trends in and implications for MS 905.5 Summary and conclusions 97References 996 Theoretical comparison of discrete-event simulation and system dynamics 105Sally Brailsford6.1 Introduction 1056.2 System dynamics 1066.3 Discrete-event simulation 1086.4 Summary: The basic differences 1106.5 Example: Modelling emergency care in Nottingham 1126.5.1 Background 1126.5.2 The ECOD project 1136.5.3 Choice of modelling approach 1146.5.4 Quantitative phase 1146.5.5 Model validation 1166.5.6 Scenario testing and model results 1166.5.7 The ED model 1186.5.8 Discussion 1196.6 The $64 000 question: Which to choose? 1206.7 Conclusion 123References 1237 Models as interfaces 125Steffen Bayer, Tim Bolt, Sally Brailsford and Maria Kapsali7.1 Introduction: Models at the interfaces or models as interfaces 1257.2 The social roles of simulation 1267.3 The modelling process 1297.4 The modelling approach 1317.5 Two case studies of modelling projects 1347.6 Summary and conclusions 137References 1388 An empirical study comparing model development in discrete-event simulation and system dynamics 140Antuela Tako and Stewart Robinson8.1 Introduction 1408.2 Existing work comparing DES and SD modelling 1428.2.1 DES and SD model development process 1438.2.2 Summary 1468.3 The study 1468.3.1 The case study 1468.3.2 Verbal protocol analysis 1478.3.3 The VPA sessions 1498.3.4 The subjects 1498.3.5 The coding process 1508.4 Study results 1518.4.1 Attention paid to modelling topics 1528.4.2 The sequence of modelling stages 1548.4.3 Pattern of iterations among topics 1558.5 Observations from the DES and SD expert modellers’ behaviour 1588.6 Conclusions 160Acknowledgements 162References 1629 Explaining puzzling dynamics: A comparison of system dynamics and discrete-event simulation 165John Morecroft and Stewart Robinson9.1 Introduction 1659.2 Existing comparisons of SD and DES 1669.3 Research focus 1699.4 Erratic fisheries – chance, destiny and limited foresight 1709.5 Structure and behaviour in fisheries: A comparison of SD and DES models 1739.5.1 Alternative models of a natural fishery 1749.5.2 Alternative models of a simple harvested fishery 1789.5.3 Alternative models of a harvested fishery with endogenous ship purchasing 1849.6 Summary of findings 1929.7 Limitations of the study 1939.8 SD or DES? 194Acknowledgements 196References 19610 DES view on simulation modelling: SIMUL8 199Mark Elder10.1 Introduction 19910.2 How software fits into the project 20010.3 Building a DES 20210.4 Getting the right results from a DES 20810.4.1 Verification and validation 21010.4.2 Replications 21110.5 What happens after the results? 21210.6 What else does DES software do and why? 21210.7 What next for DES software? 213References 21411 Vensim and the development of system dynamics 215Lee Jones11.1 Introduction 21511.2 Coping with complexity: The need for system dynamics 21611.3 Complexity arms race 21911.4 The move to user-led innovation 22111.5 Software support 22211.5.1 Apples and oranges (basic model testing) 22311.5.2 Confidence 22411.5.3 Helping the practitioner do more 23711.6 The future for SD software 24511.6.1 Innovation 24511.6.2 Communication 245References 24712 Multi-method modeling: AnyLogic 248Andrei Borshchev12.1 Architectures 24912.1.1 The choice of model architecture and methods 25112.2 Technical aspect of combining modeling methods 25212.2.1 System dynamics ® discrete elements 25212.2.2 Discrete elements ® system dynamics 25312.2.3 Agent based « discrete event 25512.3 Example: Consumer market and supply chain 25712.3.1 The supply chain model 25712.3.2 The market model 25812.3.3 Linking the DE and the SD parts 25912.3.4 The inventory policy 26012.4 Example: Epidemic and clinic 26212.4.1 The epidemic model 26212.4.2 The clinic model and the integration of methods 26412.5 Example: Product portfolio and investment policy 26712.5.1 Assumptions 26812.5.2 The model architecture 27012.5.3 The agent product and agent population portfolio 27112.5.4 The investment policy 27412.5.5 Closing the loop and implementing launch of new products 27512.5.6 Completing the investment policy 27712.6 Discussion 278References 27913 Multiscale modelling for public health management: A practical guide 280Rosemarie Sadsad and Geoff McDonnell13.1 Introduction 28013.2 Background 28113.3 Multilevel system theories and methodologies 28113.4 Multiscale simulation modelling and management 28313.5 Discussion 28913.6 Conclusion 290References 29014 Hybrid modelling case studies 295Rosemarie Sadsad, Geoff McDonnell, Joe Viana, Shivam M. Desai, Paul Harper and Sally Brailsford14.1 Introduction 29514.2 A multilevel model of MRSA endemicity and its control in hospitals 29614.2.1 Introduction 29614.2.2 Method 29614.2.3 Results 29714.2.4 Conclusion 30214.3 Chlamydia composite model 30214.3.1 Introduction 30214.3.2 Chlamydia 30214.3.3 DES model of a GUM department 30314.3.4 SD model of chlamydia 30414.3.5 Why combine the models 30414.3.6 How the models were combined 30514.3.7 Experiments with the composite model 30514.3.8 Conclusions 30714.4 A hybrid model for social care services operations 30814.4.1 Introduction 30814.4.2 Population model 30814.4.3 Model construction 30914.4.4 Contact centre model 31014.4.5 Hybrid model 31114.4.6 Conclusions and lessons learnt 313References 31615 The ways forward: A personal view of system dynamics and discrete-event simulation 318Michael Pidd15.1 Genesis 31815.2 Computer simulation in management science 31915.3 The effect of developments in computing 32015.4 The importance of process 32415.5 My own comparison of the simulation approaches 32415.5.1 Time handling 32415.5.2 Stochastic and deterministic elements 32615.5.3 Discrete entities versus continuous variables 32715.6 Linking system dynamics and discrete-event simulation 32815.7 The importance of intended model use 32915.7.1 Decision automation 33015.7.2 Routine decision support 33115.7.3 System investigation and improvement 33115.7.4 Providing insights for debate 33215.8 The future? 33315.8.1 Use of both methods will continue to grow 33315.8.2 Developments in computing will continue to have an effect 33415.8.3 Process really matters 335References 335Index 337