Date of Award


Document Type


Degree Name

Master of Science


Department of Operational Sciences

First Advisor

W. Paul Murdock, PhD

Second Advisor

Thomas F. Reid, PhD


In studying complex queueing networks, one generally seeks to employ exact analytic solutions to reduce burden on computational resources. Barring the existence of an exact solution, the alternatives include approximation techniques and simulation. Approximation is the more attractive alternative from a time and effort perspective; however, cases exist that are not amiable to this technique. This work increases the flexibility of approximation techniques in obtaining estimates of congestion measures for complex, open queueing networks having several customer classes and class-dependent structures. We accomplish this by providing the procedure to aggregate multiple classes into a single class in order to apply an existing approximation technique. The resulting method is shown to yield good agreement with results obtained by simulation. The immediate application of this work is as a tool to focus a simulation study of multi-class queueing networks. For large networks, reasonable performance estimates can be obtained quickly. Once the basic input parameters are determined, different scenarios may be rapidly evaluated in a fraction of the time needed to modify a typical simulation model. This allows one to check ideas and determine where to invest time and funding when constructing a simulation model to obtain performance estimates on a by-class basis.

AFIT Designator


DTIC Accession Number



Co-advised thesis.
The author’s Vita page is omitted.