Simulation Model Vefication Techniques

[mbrown12]

How do I verify a simulation model? Are there any specific techniques?

[rlote]

It is always advisable to verify your model before you run the experiments. Here are few techniques that can be used to verify simulation model:
A. Use modular programming concepts
B. Let only one entity Into model then watch where it goes and how long it takes to pass through the system.
C. Remove all randomness and make sure that the results match common sense and pocket-calculator arithmetic.
D. Double-check the array subscripts.
E. If it “can’t” happen, error-trap it.
F. Check that entities reach all parts of the model
G. Also make sure that the time units and distance units are consistent throughout the model.

[hoangdung]

In my opinion, when you develop simulation model you must define what is the behaviors of system which you will simulate.

And then, you will collect data from your system to validate with data you collect at real system.

For example:
You simulate a queueing system.
Data collect from simulation model are waitting time, number of customer in queue, number of customers were served by your system.
i choose above data because it is avaiable in real system. it is very easy to validate by using Statistical hypothesis testing!

[ewilliams]

Very good points. The things to be specifically simulated and measured are often called "performance metrics" in industry. For example, "average time in queue," "maximum time in queue," and "% of customers who wait > 5 minutes in queue" are all performance metrics.