SAS Simulation Studio is a SAS application that uses discrete-event simulation to model and analyze systems. Simulation Studio is based on the Java programming language and provides the following user interfaces:
a graphical user interface that requires no programming and provides all the tools for building, executing, and analyzing discrete-event simulation models
a programmatic interface that enables you to run models in batch mode
Although having a comprehensive set of modeling tools is an important quality in a simulation package, having advanced analysis tools is arguably just as important. As mentioned in the previous section, analyzing output from discrete-event simulations often requires advanced statistical methods. Simulation Studio is designed to interact with both SAS software and JMP® software so that you can conduct sophisticated statistical analyses of your results. Data generated by the model can be saved as a SAS data set or JMP table for later analysis, or alternatively you can use a SAS block included in the basic template of modeling blocks to execute SAS or JMP code directly from Simulation Studio.
Simulation Studio includes a state-of-the-art Experiment window that gives you an organized way to investigate the effects of different parameters on your model output in addition to a place to record results. For a discrete-event simulation model in general, you might be interested in conducting the following types of experiments:
a sensitivity analysis in which you vary a parameter in the model and you examine the effect on some recorded response. For example, you might be interested in the effect on customer waiting times of hiring an additional cashier at a store.
a comparison of two or more systems. For example, given two different factory floor layout options, you might want to determine which one yields a higher throughput.
an experimental design for a system that has flexibility in how several different parameters can be set. You might want to use an experimental design (such as a full factorial) to efficiently organize the testing of different parameter combinations and then study the effect on one or more results.
The Simulation Studio Experiment window can be used to conduct all these different types of simulation experiments. It can interface with JMP software to generate experimental designs and then seamlessly pass the simulated results from the design back to the JMP program for analysis. Simulation Studio is also designed to support multiple models and experiments in a single project so that you can define factors and responses once and use them for all models in the project. This is especially useful when you compare two or more systems.
No matter how advanced the available output analysis tools, they are essentially useless if you have not correctly estimated the inputs to the model. Input analysis is another important aspect of building a simulation model. In Simulation Studio input analysis can be facilitated by using JMP distribution estimation capabilities.
Simulation Studio is a flexible discrete-event simulation tool designed to provide the necessary modeling and analysis tools for both novice and advanced simulation users. Furthermore, Simulation Studio attempts to avoid being simply a black box that takes model inputs and mysteriously produces model outputs. Rather, it includes features that enable you to customize your models and tailor Simulation Studio to meet your specific needs.