A cost-driven partitioning algorithm for tandem trip-based material handling systems.

Abstract

The tandem concept for material handling systems is an approach based on the "divide and conquer" principle where a set of pick up/deposit (P/D) points is partitioned into single-vehicle, non-overlapping zones. In this dissertation, using AGV systems as an application, we present a partitioning algorithm to obtain optimal or near optimal tandem configurations for trip-based material handling systems. The objective of the partitioning algorithm is to obtain a (feasible) tandem configuration that minimizes the total system cost while satisfying user-defined performance requirements. The cost model we use is a realistic model that was developed jointly with a major AGV vendor. To measure performance, we develop an analytical model (i.e., the "WIP model") to estimate the expected WIP in each zone. Unlike previous partitioning algorithms which relied on the evaluation of individual zones, here, we evaluate each tandem configuration as a whole after the workstations (and the transfer stations) have been assigned to their appropriate zones and transit loads, if any, have been identified and routed. We also perform an empirical comparison of tandem AGV configurations with their conventional counterparts and show that tandem AGV systems are indeed a low-cost, viable alternative to conventional AGV systems. Finally, we demonstrate through an interactive design tool that a tandem configuration obtained by the partitioning algorithm is subject to change and may be further refined or improved by the user.