The model defined any clients
in the system as a variable denoted as WIP (Work in Process),
so all clients must finish traveling through the system before each
replication is complete. Both terminating conditions must be met
for each replication to finish. Each experiment ran for one hundred
replications to reduce variance and have reliable averages in the
output. Computation time was not a major issue in the experimentation,
so we chose a large number of replications for the experimentation.
This gave reasonably accurate results; for example,
with 100 replications, we obtained a 95% confidence interval for
average main lobby receptionist utilization for both the original
model as well as the kiosk model.