Network of queues and decomposition methods for modeling manufacturing systems

This research uses existing theory to develop models of flexible manufacturing systems (FMS).

Among the material handling devices most commonly used in these manufacturing systems are: microload automated storage and retrieval (AS/R) systems, automated guided vehicle systems (AGVSs), loop conveyors, transporter systems, and carousels for assembly. However, most of the models developed in this research are for the specific case of a microload AS/R system.

A network of queues is presented for the type of manufacturing systems under study. An ideal solution for this model would be an exact equilibrium distribution. However, it is shown that such a solution cannot be obtained because this model violates all known sufficient conditions for the existence of a closed form expression.

An approximate solution for the network of queues model is obtained by modifying one of the model assumptions. This approximation is called the ZTR Network. To obtain the equilibrium distribution of the ZTR Network a computational algorithm is implemented.

The ZTR Network assumes as known the expected service time at the material handling device, output buffers, and storage zones. Therefore, procedures are developed for estimating all of these expected service times for the specific case of a microload AS/RS.

Two additional models are developed for the microload AS/R system which are alternatives to the ZTR Network.

A computational study is presented that involves the comparison of the models developed in this dissertation with existing analytic methods and simulation. (Abstract shortened with permission of author.)