An Analytic Model for Design and Analysis of Single-Vehicle Asynchronous Material Handling Systems

Abstract

Most research to date on asynchronous material handling systems such as automated guided vehicle systems (AGVSs) has involved simulation studies. In this paper we develop an optimal design model for a single-vehicle AGVS that analytically accounts for the stochastic nature of real manufacturing systems. The model determines which stations and routes to include in the network to maximize the benefit of the network subject to a constraint that the average waiting time in the system not exceed a prescribed limit. In the absence of an analytical model, one would have to simulate all possible combinations of stations and routes in order to find the configuration that maximizes the benefit of the AGVS. Our contribution is to solve the problem analytically. We develop analytic expressions for key performance measures such as the full and empty trip distribution, vehicle utilization, and expected waiting time in the network. The model is formulated as a linear binary integer program with one nonlinear constraint. To solve the problem we develop bounds on the expected waiting time, which we use as part of an exact implicit enumeration solution procedure. We illustrate the model with an example of an AGVS design problem at Hewlett-Packard and we present computational experience for other representative problems.