An Analytic Model for Design of a Multivehicle Automated Guided Vehicle System

• Published in: Management science Vol. 39; no. 12; pp. 1477 - 1489
• Main Authors: Johnson, M. Eric, Brandeau, Margaret L
• Format: Journal Article
• Language: English
• Published: Linthicum, MD INFORMS 01.12.1993; Institute of Management Sciences; Institute for Operations Research and the Management Sciences
• Series: Management Science
• Subjects: AGVS; Algorithms; Analysis; Analytical models; Applied sciences; Approximation; automated systems; Automation; Computer science; control theory; systems; Control theory. Systems; Design analysis; Exact sciences and technology; Intelligent vehicles; Manufacturing; Materials handling; Mathematical models; Motor vehicle traffic; network models; Optimal solutions; Process control. Computer integrated manufacturing; Queuing theory; Systems design; Transport; Transportation; Travel time; Vehicles; Integer programming; Automation; Sensitivity analysis; Branch and bound method; Implicit enumeration method; Manufacturing; Automated guided vehicle; Transport; Queue; Handling
• ISSN: 0025-1909; 1526-5501
• DOI: 10.1287/mnsc.39.12.1477

We consider the problem of designing a multivehicle automated guided vehicle system (AGVS) to supplement an existing nonautomated material handling system. The AGVS consists of a pool of vehicles that deliver raw components from a central storage area to workcenters throughout the factor floor. The objective is to determine which workcenters warrant automated component delivery and the number of vehicles required to service those workcenters, to maximize the benefit of the AGVS, subject to a constraint that the average waiting time for material transport in the system not exceed a predefined limit. The pool of vehicles is modeled as an M / G / c queuing system and the design model is formulated as a binary integer program with nonlinear waiting time constraints, which are expressed by approximate queueing formula. We develop two different implicit enumeration algorithms to exactly solve the analytical model. We illustrate our model with an example of an actual AGVS design problem at Hewlett-Packard, and we present computational experience for other example design problems. We show how sensitivity analysis can be used to ensure that the analytical model yields an optimal solution to the design problem.