A large part of the system dynamics modeling process involves the application of common sense to dynamic problems. A good system dynamic modeler is always on the look-out for model behaviors that do not make sense. Such behaviors usually indicate a flaw in a model, and the flaw is often that a crucial piece of system structure has been left out of the model.
A common error that novice system dynamicists make is to build models with stocks whose values can either go negative, or run off to infinity. Common sense, of course, dictates that no real system can grow infinitely large, and hence that no model of a real system should be able to grow infinitely large. Similarly, common sense suggests that, since many real world variables cannot take on negative values, their model-based counterparts should not be able to take on negative values.
When a system dynamicist looks for relationships in an actual system that prevent its stocks from going negative or growing infinitely large, he or she is usually looking for the system's nonlinearities. Nonlinear relationships usually define a system's limits.
Nonlinear relationships play another important role in both actual systems and system dynamics modeling. Frequently, a system's feedback loops will be joined together in nonlinear relationships. These nonlinear "couplings" can cause the dominance of a system's feedback loops to change endogenously. That is, over time, a system whose behavior is being determined by a particular feedback loop, or set of loops, can (sometimes suddenly) endogenously switch to a behavior determined by another loop or set of loops. This particular characteristic of nonlinear feedback systems is partially responsible for their complex, and hard-to-understand, behavior. As a result, system dynamics modeling involves the identification, mapping-out, and simulation of a system's stocks, flows, feedback loops, and nonlinearities.