SAS/OR®: Introducing SAS® Simulation Studio

Discrete event simulation is used to model, study, plan, and improve systems in which random events play a dominant role. These systems are often driven by complicated mathematical and logical relationships, making it impossible to derive an analytical solution. The journal OR/MS Today cites simulation software as one of the most widely used tools in operations research, due in part to its applicability in a wide variety of fields including manufacturing, customer service, and health care. SAS Simulation Studio, experimental in SAS/OR 9.2, is a new Java-based application for modeling and analyzing systems through the use of discrete event simulation. Its graphical user interface requires no programming and provides a full set of tools for building, executing, and analyzing the results of discrete event simulation models. SAS Simulation Studio is designed to integrate with both SAS and JMP for statistical analysis of simulation results; it also can interface with JMP to generate experimental designs. SAS Simulation Studio is a flexible discrete event simulation application that provides extensive modeling and analysis tools suitable for both novice and advanced simulation users. INTRODUCTION This paper discusses the major features and functions of SAS Simulation Studio. We begin by providing some background on the nature and benefits of discrete event simulation as an analytical modeling method, highlighting its links to other forms of analysis. Turning to SAS Simulation Studio, we explain the motivation behind its creation and explore its current feature set, including the elements of the graphical user interface. We also consider the interactions between SAS Simulation Studio and JMP and the mutual benefits that result. In the appendix we walk through creation, execution, and statistics collection for a sample discrete event simulation model with SAS Simulation Studio. WHAT IS DISCRETE EVENT SIMULATION? MODELING THE REAL WORLD Simulation can be a wide-ranging term, applied in many industries and referring to many different forms of analysis. In perhaps its broadest sense, simulation denotes nothing more specific than the process of building a physical or logical model that mimics the behavior of a real-world system of interest. Physical models include wind tunnels and earthquake simulators, and a flight simulator combines physical and logical modeling. Here, however, we focus on purely logical models. Often, by using statistical analysis and related methods, it is possible to uncover logical and mathematical relationships between the key elements of a system. Furthermore, sometimes it is possible to derive an analytical solution to the problem of how to make choices that result in the best possible system performance. In many cases, however, a great number of simplifying assumptions must be made in order to create a usable analytical model and obtain a precise solution. The collective effect of these simplifying assumptions might be the creation of a model that is too simple to be a realistic depiction of the original system. Moreover, many real-world systems include not only complex mathematical and logical relationships but also significant random components. For such systems an analytical model, even a simple one, might not be possible. A far better approach is to incorporate the random elements of the system in the model. Discrete event simulation is one such modeling technique. A prime motivation for building and running a discrete event simulation model is the creation of realistic data on the performance of the system being modeled, with an overall goal of using the data to make statistically valid inferences about the performance of the system. Discrete event simulation offers the chance to generate such data without the 1 SAS Presents SAS Global Forum 2008