John S. Carson

A Sequential Procedure for Determining the Length of a Steady-State Simulation

A common problem faced by simulators is that of constructing a confidence interval for the steady-state mean of a stochastic process. We have reviewed the existing procedures for this problem and found that all but one either produce confidence intervals with coverages which may be considerably lower than desired or have not been adequately tested. Thus, in many cases simulators will have more confidence in their results than is justified. In this paper we present a new sequential procedure based on the method of batch means for constructing a confidence interval with coverage close to the desired level. The procedure has the advantage that it does not explicitly require a stochastic process to have regeneration points. Empirical results for a large number of stochastic systems indicate that the new procedure performs quite well.

Model verification and validation

In this paper we outline practical techniques and guidelines for verifying and validating simulation models. The goal of verification and validation is a model that is accurate when used to predict the performance of the real-world system that it represents, or to predict the difference in performance between two scenarios or two model configurations. The process of verifying and validating a model should also lead to improving a models credibility with decision makers. We provide examples of a number of typical situations where model developers may make inappropriate or inaccurate assumptions, and offer guidelines and techniques for carrying out verification and validation.

Introduction to modeling and simulation

Simulation is a powerful tool for the analysis of new system designs, retrofits to existing systems and proposed changes to operating rules. Conducting a valid simulation is both an art and a science. This paper provides an introduction to simulation and modeling and the main concepts underlying simulation. It discusses a number of key issues regarding a simulation team, how to conduct a simulation study, the skills required and the steps involved. It also provides project management guidelines and outlines pitfalls to avoid.Simulation is a powerful tool for the evaluation and analysis of new system designs, modifications to existing systems and proposed changes to control systems and operating rules. Conducting a valid simulation is both an art and a science. This paper provides an introduction to discrete event simulation and the main concepts - system state, events, processes - underlying simulation. It discusses the major world views used by simulation software. It includes a brief discussion of a number of other important issues: the advantages and disadvantages of using a simulation model, the skills required to develop a simulation model, the key steps in conducting a simulation study, as well as some project management guidelines and pitfalls to avoid.

Integrating optimization and simulation: research and practice

The integration of optimization and simulation has become nearly ubiquitous in practice, as most discrete-event simulation packages now include some type of optimization routine. This panel sessions objective was to explore the present state of the art in simulation optimization, prevailing issues for researchers, and future prospects for the field. The composition of the panel included views from both simulation software developers and academic researchers. This Proceedings paper begins with a brief overview of some issues, introduced by the chairman and organizer of the session, followed by the position statements of the panel members, which served as a starting point for the panel discussion.

Conservation Equations and Variance Reduction in Queueing Simulations

We consider the efficient estimation of mean delay in queue, <italic>d</italic>, mean wait in system, <italic>w</italic>, time average number in queue, <italic>Q</italic>, and time average number in system, <italic>L</italic>, for simulated queueing systems. We prove for the <italic>GI/G/s</italic> queue that it is more efficient to estimate <italic>w, Q</italic>, and <italic>L</italic> from an estimate of <italic>d</italic> than it is to estimate them directly. This generalizes previous results for the <italic>M/G/</italic>l queue and also confirms empirical studies on other <italic>GI/G/s</italic> queues.

A Note on Spira’s Algorithm for the All-Pairs Shortest-Path Problem

We correct some errors in Spira’s algorithm for the all-pairs shortest-path problem, and empirically compare his algorithm (with two distinct sorting, routines) to Dijkstra’s procedure. The results show that Spira’s algorithm is only efficient for “large” networks. Furthermore, it is seen that the asymptotic number of additions and comparisons required by two algorithms may be a very poor indicator of their relative running times.

Verification And Validation: A Consultant's Perspective

The purpose of this paper is to look at the questions and issues regarding verification and validation of simulation models from the industrial and/or consultants perspective. We will discuss some of the practical as well as conceptual issues involved. In addition, this paper and presentation will briefly describe several simulation projects/case studies and the verification and validation techniques used in them.

Introduction to discrete-event simulation

In this article, we introduce the reader to discrete-event simulation. The concepts of system and model, system state, entities, attributes and delays are defined in the general context of simulation. Using these concepts, event-scheduling, process-interaction, and activity-scanning perspectives are briefly described. To demonstrate the use of the concepts, a discrete system is modeled using the event-scheduling perspective. Simulation languages are classified in terms of the type of system being modeled, the application level, and the perspective taken. The features of a simulation language are discussed. Lastly, basic information is provided about an assortment of discrete-event simulation languages.

Modeling and simulation worldviews

This paper provides a language-independent introduction to the major simulation modeling world views for discrete-event systems simulation. A world view is the modeling framework that a modeler uses to represent a system and its behavior. The main terminology and concepts include systems and models, system state variables, entities and their attributes, lists, resources, events, activities and delays. These concepts are covered from the perspective of the modeler. Finally, in the tutorial, we will attempt to make these ideas concrete through the use of a number of examples and exercises.

AutoStat: output statistical analysis for AutoMod users

AutoStat/sup TM/ is an extension package for AutoMod/sup TM/ and AutoSched/sup TM/ models that provides complete support for simulation model experimentation and statistical analysis of outputs. Within a menu-driven, point and click environment, AutoStat provides assistance for setting up runs, automated execution of runs and consolidation of outputs across replications and scenarios. AutoStat automatically sets random number seeds to achieve statistically independent replications; it also sets factor levels (input parameters) to realize desired scenarios, without having to modify the underlying model. AutoStat has a number of advanced features such as datafile factors, responses (or performance measures) read from custom user output files, and user-defined combination responses (linear combinations of other responses). AutoStat offers several statistical methods, including confidence intervals, a ranking and selection procedure, design of experiments, and warm-up determination. Outputs can be displayed and printed in tabular and graphical formats and exported to spreadsheet and other analysis software. AutoStat saves simulation modelers considerable time and automates the effort of setting up and making runs, managing the input and output files, consolidating results and conducting analyses.