Page 691 - ExtendSim User Guide
P. 691

Technique Advantages
Sensitivity Analysis Easiest to specify
Included in all ExtendSim products
Scenario Manager
Gives the most complete understanding of how the model reacts to different factors
Factors can be from data- bases as well as dialog variables
Automatically records each response for every simulation run
Design of experiments can reduce the number of scenarios required to study factor interactions
A set of scenarios must be created by the modeler
May require many simu- lation runs to evaluate all of the scenarios
Only available in Extend- Sim AT and ExtendSim Suite
“Scenario analysis” on page 650.
Analysis 665
Determines the optimum model configuration
Constraints can be added to filter out infeasible model configurations
Included in all ExtendSim products
An objective function is required
Does not record the results of each individual run
There is a practical limit on the number of factors, and their possible values, that can be examined
May require many simu- lation runs to determine the optimal configuration
“Optimization” on page 665.
Disadvan- tages
Evaluating more than one factor is cumbersome
Additional modeling is required to record model results
“Sensitivity analysis” on page 646.
Refer to
Optimization is a powerful feature that can automatically determine ideal values for parame- ters in a model. It does this by running the model many times using different values for selected parameters, searching the solution space until it is satisfied that it has found an accept- able solution. It then populates the model with the optimized parameter values.
ExtendSim facilitates optimization by making the optimization algorithm available within a block that can be added to any model to control all aspects of the optimization. Furthermore, having a block do the optimization increases flexibility and opens up the method and source code to users who might want to modify or create their own customized optimization blocks. The Optimizer block (Value library) uses an evolutionary algorithm to reduce the number of times the model has to run before a solution is found.
How optimization works
Optimization, sometimes known as “goal seeking,” is a useful technique to automatically find the best answer to a problem. The “problem” is stated as an objective function or cost equation
How To

   689   690   691   692   693