Simulation: The Practice of Model Development and Use
February 2004, ©2004
- How a simulation works
- Selecting simulation software
- Designing an appropriate conceptual model
- Developing the computer based model
- Experimenting with the simulation
- Verifying and validating the simulation
- Simulation in practice
Two case study examples illustrate the principles that are described. Simulation modellers need to apply these principles whatever the software they are using for developing models.
CHAPTER 1: www.simulation: What, Why and When?
1.2 What is simulation?
1.3 Why simulate?
1.3.1 The nature of operations systems: variability, interconnectedness and complexity.
1.3.2 The advantages of simulation.
1.3.3 The disadvantages of simulation.
1.4 When to simulate.
CHAPTER 2: Inside Simulation Software.
2.2 Modelling the progress of time.
2.2.1 The time-slicing approach.
2.2.2 The discrete-event simulation approach (three-phase method).
2.2.3 The continuous simulation approach.
2.2.4 Summary: modelling the progress of time.
2.3 Modelling variability.
2.3.1 Modelling unpredictable variability.
2.3.2 Random numbers.
2.3.3 Relating random numbers to variability in a simulation.
2.3.4 Modelling variability in times.
2.3.5 Sampling from standard statistical distributions.
2.3.6 Computer generated random numbers.
2.3.7 Modelling predictable variability.
2.3.8 Summary on modelling variability.
CHAPTER 3: Software for Simulation.
3.2 Visual interactive simulation.
3.3 Simulation software.
3.3.2 Programming languages.
3.3.3 Specialist simulation software.
3.3.4 Comparing spreadsheets, programming languages and specialist simulation software.
3.4 Selection of simulation software.
3.4.1 The process of software selection.
3.4.2 Step 1: Establish the modelling requirements.
3.4.3 Step 2: Survey and shortlist the software.
3.4.4 Step 3: Establish evaluation criteria.
3.4.5 Step 4: Evaluate the software in relation to the criteria.
3.4.6 Step 5: Software selection.
CHAPTER 4: Simulation Studies: An Overview.
4.2 Simulation studies: an overview of key modelling processes.
4.2.1 Simulation modelling is not linear.
4.2.2 Something is missing!
4.3 Simulation project time-scales.
4.4 The simulation project team.
4.5 Hardware and software requirements.
4.6 Project costs.
4.7 Project selection.
CHAPTER 5: Conceptual Modelling.
5.2 Conceptual modelling: important but little understood.
5.3 What is a conceptual model?
5.4 Requirements of the conceptual model.
5.4.1 Four requirements of a conceptual model.
5.4.2 Keep the model simple.
5.5 Communicating the conceptual model.
5.5.1 Simulation project specification.
5.5.2 Representing the conceptual model.
CHAPTER 6: Developing the Conceptual Model.
6.2 A Framework for conceptual modelling.
6.2.1 Developing an understanding of the problem situation.
6.2.2 Determining the modelling objectives.
6.2.3 Designing the conceptual model: the inputs and outputs.
6.2.4 Designing the conceptual model: the model content.
6.2.5 The role of data in conceptual modelling.
6.2.6 Summary of the conceptual modelling framework.
6.3 Methods of model simplification.
6.3.1 Aggregation of model components.
6.3.2 Excluding components and details.
6.3.3 Replacing components with random variables.
6.3.4 Excluding infrequent events.
6.3.5 Reducing the rule set.
6.3.6 Splitting models.
6.3.7 What is a good simplification?
CHAPTER 7: Data Collection and Analysis.
7.2 Data requirements.
7.3 Obtaining data.
7.3.1 Dealing with unobtainable (category C) data.
7.3.2 Data accuracy.
7.3.3 Data format.
7.4 Representing unpredictable variability.
7.4.2 Empirical distributions.
7.4.3 Statistical distributions.
7.4.4 Traces versus empirical distributions versus statistical distributions.
7.4.6 Further issues in representing unpredictable variability: correlation and non-stationary data.
7.5 Selecting statistical distributions.
7.5.1 Selecting distributions from known properties of the process.
7.5.2 Fitting statistical distributions to empirical data.
CHAPTER 8: Model Coding.
8.2 Structuring the model.
8.3 Coding the model
8.3.1 Separate the data from the code from the results.
8.3.2 Use of pseudo random number streams.
8.4 Documenting the model and the simulation project.
CHAPTER 9: Experimentation: Obtaining Accurate Results.
9.2 The nature of simulation models and simulation output.
9.2.1 Terminating and non-terminating simulations.
9.2.2 Transient output.
9.2.3 Steady-state output.
9.2.4 Other types of output.
9.2.5 Determining the nature of the simulation output.
9.3 Issues in obtaining accurate simulation results.
9.3.1 Initialization bias: warm-up and initial conditions.
9.3.2 Obtaining sufficient output data: long runs and multiple replications.
9.4 An example model: computer user help desk.
9.5 Dealing with initialization bias: warm-up and initial conditions.
9.5.1 Determining the warm-up period.
9.5.2 Setting initial conditions.
9.5.3 Mixed initial conditions and warm-up.
9.5.4 Initial conditions versus warm-up.
9.6 Selecting the number of replications and run-length.
9.6.1 Performing multiple replications.
9.6.2 Variance reduction (antithetic variates).
9.6.3 Performing a single long run.
9.6.4 Multiple replications versus long runs.
CHAPTER 10: Experimentation: Searching the Solution Space.
10.2 The nature of simulation experimentation.
10.2.1 Interactive and batch experimentation.
10.2.2 Comparing alternatives and search experimentation.
10.3 Analysis of results from a single scenario.
10.3.1 Point estimates.
10.3.2 Measures of variability.
10.4 Comparing alternatives.
10.4.1 Comparison of two scenarios.
10.4.2 Comparison of many scenarios.
10.4.3 Choosing the best scenario(s).
10.5 Search experimentation.
10.5.1 Informal approaches to search experimentation.
10.5.2 Experimental design.
10.5.4 Optimization ("searchization").
10.6 Sensitivity analysis.
CHAPTER 11: Implementation.
11.2 What is implementation?
11.2.1 Implementing the findings.
11.2.2 Implementing the model.
11.2.3 Implementation as learning.
11.3 Implementation and simulation project success.
11.3.1 What is simulation project success?
11.3.2 How is success achieved?
11.3.3 How is success measured?
CHAPTER 12: Verification, Validation and Confidence.
12.2 What is verification and validation?
12.3 The difficulties of verification and validation<.
12.3.1 There is no such thing as general validity.
12.3.2 There may be no real world to compare against.
12.3.3 Which real world?
12.3.4 Often the real world data are inaccurate.
12.3.5 There is not enough time to verify and validate everything.
12.3.6 Confidence not validity.
12.4 Methods of verification and validation.
12.4.1 Conceptual model validation.
12.4.2 Data validation.
12.4.3 Verification and white-box validation.
12.4.4 Black-box validation.
12.4.5 Experimentation validation.
12.4.6 Solution validation.
12.5 Independent verification and validation.
CHAPTER 13: The Practice of Simulation.
13.2 Types of simulation model.
13.3 Modes of simulation practice.
13.3.1 Three modes of practice.
13.3.2 Facets of the modes of simulation practice.
13.3.3 Modes of practice in business and the military.
APPENDIX 1: Wardeon Cinema.
APPENDIX 2: Panorama Televisions.
APPENDIX 3: Methods of reporting simulation results.
APPENDIX 4: Statistical distributions.
APPENDIX 5: Critical values for the chi-square test.
APPENDIX 6: Critical values for the Student’s t-distribution.
- Develops a step-by-step guide for managing simulation projects, providing the reader with a clear understanding of the requirements for the successful development and use of simulation models
- Follows a non-software-specific approach with a focus on solving problems with discrete event simulation - involves limited mathematical content, with clear explanations of key statistical concepts
- Focused on real-world application - key concepts are illustrated by two case examples that run throughout the book, one based on a service call centre model and the other on a manufacturing assembly line
- Involves limited mathematical content, with clear explanations of key statistical concepts
"The success of a simulation study depends on the adopted simulation tools, use and understanding of statistical methods and adequate management skills. The latter skills refer to the integrative use of tools and statistics within a project frame. This book addresses just these skills, building on the latest scientific insights and as such is a significant help for practitioners and students in mastering their simulation studies and providing adequate answers to the questions posed." Durk-Jouke van der Zee, Assistant Professor, Production Systems Design Group, Faculty of Management and Organization, University of Groningen