Industrial Science makes a good point about why we need simulation models and how, conceptually, the process of building a useful model works.
I like to summarize the process of creating and using models into roughly three activities. First, frame the problem by identifying the behaviors of interest, choosing system boundaries that will capture these behaviors and stating a structural hypothesis about how the system works to create the behaviors of interest.
Second, build a simulation model to test the hypothesis.
Finally, use the model to find management policies that improve system behavior..
Obviously, in practice this entire process is a bit iterative (you know: rinse, repeat).
In practice, applying complexity science means writing and using simulation models.
