Lesson 10 - Unit 14 Summary

The primary objective of building simulation models is to better understand the controlling factors and processes in the system and use this information to predict (forecast) the future behavior of the system under various sets of assumptions. The primary reason we would do so is so we can determine what we can do to influence the future behavior of the system.  That is, our goal is often to use the simulation model to identify any necessary changes we can make to the system that will help make it perform the way that we want it to. Such changes range from changing the physical design of the system (e.g., increasing the number of scanning stations at a security checkpoint) to changing policies for how the system is run (e.g., changing the decision rules for opening and closing new stations).

To accomplish this in GoldSim (or any simulation tool), what we want to do is: 1) build a simulation model; 2) run it multiple times using different inputs (representing different designs or policies); and then 3) compare the results of those different simulation runs to determine which set of inputs best meets our goals for the system.

To support this, GoldSim provides a powerful and flexible feature referred to as scenario modeling. GoldSim’s scenario modeling capability allows you to directly compare results generated by your model using different sets of input parameters.  In effect, when you use this capability, your model can store (and subsequently compare) multiple sets of inputs and outputs. This Unit discussed this important feature.

In particular, the key points that we covered were as follows:

• In GoldSim, a scenario is a specific set of input data (and corresponding outputs) for a model. When modeling scenarios, multiple scenarios will be defined within a model.  Different scenarios within a model are specifically differentiated by having different values for one or more Data elements.  These elements are referred to as Scenario Data.
• The only elements that can differ between scenarios in a model are Data elements. Of course, not all Data elements in a model will actually differ between scenarios.  In most cases (i.e., anything other than very simple models), you will define a subset of the model’s Data elements as being Scenario Data.
• Scenario Data elements can be used to differentiate scenarios in complex ways, such as having different scenarios use different time series inputs, different Lookup Tables, or even completely different logic for some parts of the model.
• You can view a list of defined scenarios in a model in the Scenario Manager. The Scenario Manager can be accessed from the main menu (Run|Scenario Manager…) or by pressing F7.
• There are two ways to edit/create Scenario Data: 1) You can do so directly in the Scenario Data dialog accessed from the Scenario Manager; or 2) you can edit Data elements directly for the Active Scenario.
• The Active Scenario is the scenario that is being viewed when you are browsing a model. That is, whenever a model has scenarios, one of the scenarios is the Active Scenario.  The Active Scenario is indicated in the lower right-hand corner of the GoldSim status bar.
• In order to compare scenarios, the model must be in Scenario Mode.
• When in Scenario Mode, scenario results can only be viewed in Time History Result elements, Final Value Result elements and Distribution Result elements. No other results are available.  Hence, if your model does not contain any Time History, Final Value or Distribution Result elements, you will not be able to compare scenario results.
• By definition, a model can only be in Scenario Mode if it has at least one scenario with results.  However, when in Scenario Mode, not all scenarios necessarily have results (since the scenarios can be run independently).
• You enter Scenario Mode by running one or more scenarios from the Scenario Manager, or by choosing to enter Scenario Mode from Result Mode.

We also illustrated that Time Series elements have an advanced feature that allows them to store multiple data sets, and this can be used to use different time series data for different scenarios.