By Gene Munce
Simulation is a method to mimic physical behaviors in a system for observability. This may be needed because a system is small and fast such as an IC. Or it may be because a system is large and spread out such as a nationwide distributor.
Sometimes a simulation is for real-time observation of a system or process. In this case a model is built to reflect the system under study. Real-time signals feed the model to cause a change in state. The model is often visualized (e.g. lines of people, or trucks in transit). Rules can be instituted to flag abnormal conditions or alerts to trigger an operations manager into action.
Sometimes a simulation is for replay and experimentation of a system or process. In this case a model or models are built to reflect the system under study or variations of it. The input to the simulation is built to reflect real world state change input. This can be a trace of actions or characterization of real-time activity. In the second case the input is described statistically (e.g. rate of arrival, distribution curves, …). Multiple simulations can be run with differing configuration options to study the effect (e.g. differing queue lengths, or service times).
In both cases bottlenecks can be observed and remedied. In the first case trucks could be rerouted to optimize delivery. In the second case changes to a design can be suggested in advance of production.