Page 626 - ExtendSim User Guide
P. 626

Model Execution
Running a model
Running a model
In the ExtendSim Tutorial you learned the basic steps in running a model. Some additional points are discussed in the following sections.
Menu commands and toolbar buttons
You can run, stop, and pause simulations using toolbar buttons or the commands under the Run menu.
• The Run menu commands are discussed at “Run menu” on page 835.
• Buttons to activate the Run commands are shown at “Toolbar buttons” on page 841.
Running a model multiple times
It is common that you will build a model and run it repeatedly. Running a model multiple times gives a range of values indicating possible outcomes and facilitates model analysis. For exam- ple, you would do this when the model has random inputs, for Monte Carlo simulations, or when performing sensitivity analysis.
• As discussed in “Constant values and random variables” on page 60, models with one or more random inputs are called stochastic models. Stochastic models are run repeatedly and then analyzed statistically to determine a likely outcome.
• Monte Carlo simulation is commonly defined as applying random behavior to a static or dynamic model and evaluating the results over a number of trials or observations. Applying Monte Carlo simulation techniques to a dynamic model is also known as stochastic model- ing, which is discussed above. You can build both static and dynamic models with Extend- Sim. For more information, see “Monte Carlo modeling” on page 49.
• When you perform sensitivity analysis on a model, you select a variable for analysis and vary it to determine how much of an impact the variable has on the model as a whole. Sensi- tivity analysis is discussed in detail in “Sensitivity analysis” on page 646.
You should plan on running a simulation at least 3-5 times to estimate the variability in output whenever a model contains random variables. Then you can use statistical sample-size calcula- tions (discussed in “Determining the length and number of runs” on page 609) to determine how many additional simulation runs are required to obtain an accurate estimate of the model’s performance. Some tools in ExtendSim for evaluating multiple simulation runs are:
• Histogram blocks (Plotter library) that show the shape and any tendencies in the range of results
• Error Bar plotters (Plotter library) that show how a variable is affected by randomness through the course of the simulation run
• MultiSimplotters(Plotterlibrary)thatcanretaindataandshowtheresultsofuptofoursim- ulation runs
• Confidenceintervals(reportedbytheStatisticsblockintheValuelibrary)thatshowthe probability that a certain range captures the true mean for a simulation model result
Stepping through a model
Sometimes you want to pause a simulation run to observe something of special interest, or step through the run to verify the action. There are several ways to accomplish this:
How To

   624   625   626   627   628