compares typical performance results for the Markovian
queueing models with and without customer abandonment. For
this illustration, t¼20 s, hi¼300 s, b¼300 s (where applicable),
si¼100, and li is parameterized to produce a range of loading levels
lihi=si between 70% and 150%. The comparatively high sensitivity
of the conventional model is noteworthy, with service level
dropping from nearly 100% at 80% loading to 0 at 100% loading.
In contrast, the improved model indicates a service level of about
70% at 100% loading with about 4% of customers abandoning. At
150% loading, the service level has decayed to nearly zero and, as
we would expect, one-third of the arriving customers abandon the
queue. Even though abandoning customers cannot themselves be
served, they significantly improve service level performance by
reducing system congestion precisely under those conditions when
the reduction is most beneficial.