Skip to main content
main-content

Über dieses Buch

Discrete event simulation and agent-based modeling are increasingly recognized as critical for diagnosing and solving process issues in complex systems. Introduction to Discrete Event Simulation and Agent-based Modeling covers the techniques needed for success in all phases of simulation projects. These include: • Definition – The reader will learn how to plan a project and communicate using a charter. • Input analysis – The reader will discover how to determine defensible sample sizes for all needed data collections. They will also learn how to fit distributions to that data. • Simulation – The reader will understand how simulation controllers work, the Monte Carlo (MC) theory behind them, modern verification and validation, and ways to speed up simulation using variation reduction techniques and other methods. • Output analysis – The reader will be able to establish simultaneous intervals on key responses and apply selection and ranking, design of experiments (DOE), and black box optimization to develop defensible improvement recommendations. • Decision support – Methods to inspire creative alternatives are presented, including lean production. Also, over one hundred solved problems are provided and two full case studies, including one on voting machines that received international attention. Introduction to Discrete Event Simulation and Agent-based Modeling demonstrates how simulation can facilitate improvements on the job and in local communities. It allows readers to competently apply technology considered key in many industries and branches of government. It is suitable for undergraduate and graduate students, as well as researchers and other professionals.

Inhaltsverzeichnis

Frontmatter

Chapter 1. Introduction

Abstract
Discrete event simulation and agent-based modeling are the subjects of this book. These types of simulation are merely two of many with others including systems dynamics, finite element analysis (FEM), and physical, human simulation. This latter type can involve running actual people through a scenario or game. Yet, discrete event simulation and agent-based modeling can offer natural approaches to help people think introspectively about their systems and realize efficiency gains. An animated version of the system being modeled is often a major outcome of the modeling activity. These animations can provide the best available way to engage untrained people in the application of all types of operations research and systems improvement activities.
Theodore T. Allen

Chapter 2. Probability Theory and Monte Carlo

Abstract
Queried to come from Author.
Theodore T. Allen

Chapter 3. Input Analysis

Abstract
This chapter describes methods for gathering and analyzing real world data to support discrete event simulation modeling. In this phase, the distribution approximations for each process including arrivals are estimated using a combination of field observations (e.g., based on stopwatch timings) and assumption-making. In many cases, the time and cost of input analysis will actually exceed expenses from all other phases. It may also be necessary to put instrumentation into place to provide the accurate time measures, greatly delaying the entire project.
Theodore T. Allen

Chapter 4. Simulating Waiting Times

Abstract
This chapter describes the standard process for performing discrete event simulations to estimate expected waiting times. In doing so, it describes event-based “controllers” that generate chronologies differentiating discrete event simulation from other types of statistical simulations.
Theodore T. Allen

Chapter 5. Output Analysis

Abstract
After input analysis, model building, and model validation, decision support is not immediately available. The simulation team simply has a model to predict outputs or responses for given combinations of input or factor settings. Showing related animations and the results from a single system is rarely sufficient. While the process of building the model itself likely lead to insights and valuable data, using creativity to generate alternative systems to be evaluated is almost always critical to the success of the project. By single “system” we mean one combination of numbers of machines, staffers, staff schedules, and other factor levels which could, e.g., represent the current operating conditions.
Theodore T. Allen

Chapter 6. Theory of Queues

Abstract
Queuing theory models provide these benefits with at least two types of associated costs. First, users need to make a limiting set of assumptions about arrivals and service distributions. These assumptions might not apply to any reasonable approximation in some cases of interest. Making them could lead to inadvisable recommendations. Second, queuing theory models are associated with complexity and abstract concepts. These take time to understand and to apply confidently.
Theodore T. Allen

Chapter 7. Decision Support and Voting Systems Case Study

Abstract
This chapter describes practical information relevant to simulation projects. In general, the inspiration for system changes can come from many sources including from competitors. Also, the creativity involved with identifying alternative systems is generally critical to project success. Simulation is usually only useful for evaluating hypothetical changes. Without inspired alternatives, the value is limited. Dogmas like theory of constraints (Sect. 7.1) and lean production (Sect. 7.2) can provide the needed inspiration.
Theodore T. Allen

Chapter 8. Variance Reduction Techniques and Quasi-Monte Carlo

Abstract
Computer speeds continue to increase. At the same time, the complexity and realism of simulations also continues to increase. For example, 20 replicates of the voting systems simulation in Chap.​ 7 involve approximately 10 million simulated voters. Currently, a standard PC requires several minutes to yield the expected worst precinct closing time estimate. In 5 years, the identical simulation might require less than a single minute. Yet, we might choose to include in the simulation details about registration process and voter demographics so that the resulting time might require more than 10 min again.
Theodore T. Allen

Chapter 9. Simulation Software and Visual Basic

Abstract
Queried to come from Author.
Theodore T. Allen

Chapter 10. Introduction to ARENA Software

Abstarct
This chapter introduces a software package used widely for instruction and real-world decision-support often referred to as ARENA software. ARENA is effectively a suite of software which includes the ARENA simulator, the Input Analyzer (for input analysis), and the Process Analyzer (PAN) (for output analysis). All of these are produced by Rockwell International which is also a major manufacturer. This situation permits the ARENA team to constantly apply their software to real-world problems. ARENA software and the related suite do not have the full range of graphics and visualization capabilities of software such as AutoMod, MODSIM, PROMODEL, SIMIO, and many others. Also, the Input Analyzer and the PAN lack statistical analysis features in more powerful software such as SAS. Yet, ARENA is comparatively easy-to-learn, of moderate cost (around $20K per a professional license), and can be successfully used for even large projects.
Theodore T. Allen

Chapter 11. Advanced Modeling with ARENA

Abstract
In particular, the “Decide” module provides perhaps the most critically important way to build structure into simulations. Decide can route entities based on a probabilistic condition such as whether or not parts conform to specifications. More commonly, perhaps, it can route based on system conditions. For example, an entity might enter the shorter queue or be assigned to the more productive process. In the latter case, we might designate Process1 and Process2 as the two servers. Then, the entity routing condition in the Decide block could be based on the expression: Process 1.NumberOut <= Process 2.NumberOut. If the condition is true, the entity departs the decide block through the upper exit point, otherwise, the lower point.
Theodore T. Allen

Chapter 12. Agents and New Directions

Abstract
In this chapter, computer simulation approaches in addition to discrete event simulation are described. The focus is primarily on agent-based modeling which is defined as the activity of simulating system-wide properties as they derive from the actions and interactions of autonomous entities. This contrasts with system dynamics and other differential equation-based modeling including finite element methods (FEM). In FEMs, e.g., there are relatively few entities and the interplay of physical or cognitive forces dominates.
Theodore T. Allen

Chapter 13. Answers to Odd Problems

Without Abstract
Theodore T. Allen

Backmatter

Weitere Informationen

Premium Partner

    Bildnachweise