Robust Design of Automated Guided Vehicles System in an FMS

Abstract

Automated Guided Vehicles (AGV), as material handling systems, are widely diffused in FMS environment. The design of such a system involves the selection of the most suitable lay-out on one hand, and the choice of the “optimal” level of some parameters such as the number of vehicles. machine buffer capacity, the number of pallets, vehicle and part dispatching rules, etc. on the other. The optimal combination of these factor levels, that maximises a certain output variable, could be uncovered by Response Surface Methodology (RSM). But two are the problems that immediately arise in the application of such a technique:

• how to consider the qualitative variables, like dispatching or loading rules;

• how to think about some uncontrollable factors, like production mix or machine Mean Time Between Failure (MTBF), whose level can be controlled only during simulations (Noise Factors — NF).

The Japanese researcher Taguchi suggests to study the variability in the system performance induced by these NFs in order to select the best setting (the least sensitive to the induced variability) of the controllable ones (CFs).

In this research we propose a new approach to get the best solution to our problem, considering (in contrast with what Taguchi advises) a single matrix for both CFs and NFs, potentially fractionable, using different responses for the analysis of simulation results, including Taguchi’s S/N (signal to noise) ratio.