|
|
|
|
|
|
|
Figure 9.14
Asymptotic model error for low populations. |
|
|
|
| |
|
|
|
|
A 10% error in occupancy can cause significantly larger errors in queue size if the occupancy is large (e.g., r > .8). Care should be used if the asymptotic model gives a result of ra > .8. In these cases, one should consider the low population model for cases in which n < 10 or even 20. |
|
|
|
| |
|
|
|
|
Accuracy in any model depends on the validity of the user parameters. A highly accurate model with inaccurate input parameters still produces an inaccurate characterization of system behavior. More complex models are susceptible to incorrect specification of input parameters. |
|
|
|
|  |
|
|
|
|
We have tried in this chapter to choose an approximate (asymptotic) model that can be well understood by designers and can provide an initial estimate of performance across a relatively broad range of systems. This can then be the basis for initial design tradeoff studies. Later, more sophisticated simulation, or more detailed analytic studies, can provide additional system modeling detail. |
|
|
|
|
|
|
