8.2 Time control

< Free Open Study >

In continuous and discrete simulation, the major concern in performing the simulation is time management and its use in affecting the dependent variables. Timing in simulation programs is used to synchronize events, compute state changes, and control overall interactions. Timing can take on two modes: synchronous and asynchronous.

Synchronous timing refers to a timing scheme in which time advances in fixed, appropriately chosen units of time, t. On each update of time the system state is updated to reflect this time change. That is, all events occurring during this time period are determined and their state adjusted accordingly. This process of advancing time (in steps) and updating the state of elements occurs until the simulation hits some boundary condition (time goes to a maximum, some event occurs, etc.). In our bank teller system timing is needed to determine arrivals and service. For the t-step organization on each stop we must check to see if an arrival should occur, if a service should be completed, or if a new one should be begun. An important concept or idea to keep in mind when using synchronous timing is that of step selection. If too great a step is chosen, events are seen to occur concurrently when in reality they may not be. On the other hand, too fine a granularity of time step will cause many steps to go by when nothing occurs. The latter will cause excessive computer run time but very fine differentiation between events. The former, on the other hand, will cause a distortion of everything and possibly a loss of usefulness. The important job of the modeler is to select the proper step time for the model to be useful but not be excessive in computer time.

Asynchronous, or event timing, differs from synchronous timing in that time is advanced in variable rather than fixed amounts. The concept is to keep track of events versus time steps. The time is advanced based on the next chronological event that is to occur. These chronological events are typically kept in a dynamic list, which is updated and adjusted on each event to reflect new events that are to occur in the future.

In our bank teller example the event queue, or list, will comprise two events: the next arrival and the completion of the next service. Abstractly this method appears to be easier to visualize. The events must be ordered by occurrence and adjusted as new events arrive. The issue in this, as well as in the former case, is how to insert or schedule new events or new conditions in the simulations. The next section will investigate this and other aspects of how to use time in building simulations.

< Free Open Study >