| This work focuses on the development of Markov-chain models for the analysis of bulk-arrival, bulk-service queues operating under the assumption of homogeneous and non-homogenous Poisson arrivals. The theory developed here is geared towards measuring the number stranded in bulk-arrival, bulk-service queues. The number stranded is directly related to the Quality of Service (QoS) of the server in the queuing system. Examples of such queues, commonly formed in the real world, are: transportation systems with (non-unit load) automated guided vehicles in production factories and people-movers in amusement parks. The analysis in this dissertation takes into consideration various dispatching and stranding strategies encountered in the real world. The Markov-chain models developed in this work avoid the well-known, classical transformation techniques employed in queuing theory. The use of these transformation techniques can lead to severe numerical difficulties; the Markov-chain techniques presented here bypass some of these difficulties for measuring the number stranded. To the best of this researcher's knowledge, there is no literature that analyzes the number stranded in bulk-arrival, bulk-service queues.;Theoretical analysis is supplemented with a case study from technology management. In particular, the case study is related to a problem in strategic capacity management of dedicated air-cargo-delivery systems. This problem needs to be analyzed under a variety of dispatching and "stranding strategies," since these strategies vary across organizations and across cities within the same organization. Finally, extensive computational tests are performed to demonstrate the efficacy of these models. |
