Based on the discussion in this chapter and in Chapter 3, we have introduced many of the statistical aspects of simulation modeling and analysis. The first issue that you face is determining whether or not the simulation has a finite or an infinite horizon. Based on that determination, you must determine how much to sample in terms of the number of replications or in the case of batch means the length of the simulation run. We have discussed using the normal approximation method and the half-width ratio method with Chapter (ch3). In the case of an infinite horizon simulation the same concepts reappear when using the method of replication deletion or when performing a batch means analysis. After determining your statistical environment you can better plan out how you will execute the simulation to make decisions. The same techniques that were discussed in Section 4.4 of Chapter 4 can still be used when comparing systems that have an infinite horizon. If you use the method replication-deletion, then there no conceptual difference in comparing of two design alternatives. If you choose to use the batch means method, then the approach is to utilize the batch means from each of the two design configurations as if they were random samples. Unfortunately, batching complicates the use of common random numbers. Thus, I recommend utilizing the method of replication-deletion when your infinite horizon simulation scenarios must be compared.
The next chapter will address more advanced modeling techniques for process modeling. Specifically, non-stationary arrivals, additional concepts in resource modeling, and other miscellaneous modeling techniques. At this point, you should be able to model a wide variety of interesting systems and perform the basic statistical analysis on those models that are required to make valid statistical conclusions.