Combined effect of job and AGV dispatching rules on a flexible manufacturing system
In a Flexible Manufacturing System (FMS) jobs are moved between work-centers by an Automated Guided Vehicle System (AGVS). The operation of the AGVS is governed by two kinds of AGV dispatching rules: Workcenter Initiated Rules (WI), and Vehicle Initiated Rules (VI). At work-centers. jobs are prioritized for processing by Job Dispatching (JD) Rules.
The objective of the study is to characterize system behavior when dispatching rules from each of the three categories are used concurrently -- an issue which has received very little attention in published research. In addition to common measures of system performance such as job flowtime and shop throughput, measures specifically directed at AGVS effectiveness such as waiting times and queue lengths for AGVs, and empty AGV travel are used. Combinations of rules from the three categories are studied under conditions of high and low AGV utilization.
The simulation model of the FMS was developed using SLAM II - Material Handling Extension. The hypothetical FMS consists of four non-identical workcenters which process multiple part types. A fleet of four AGVs move the jobs between workcenters.
Results indicate that for the system analyzed, WI rules (for vehicle selection) do not play a significant role in determining system performance even when the AGV fleet is under-utilized. For the system analyzed, V1 rules (for selecting jobs to picked up by AGVs) were found to have a significant effect on all measures of performance, only under conditions of constrained AGV fleet size. Good results were obtained with the Shortest Travel Distance rule.
The relative performance of VI rules was found to be independent of job dispatching rules. However, some combinations of these rules interacted to produce inexplicable results. A notable observation of this research was that job dispatching rules have a significant effect on measures of AGVS performance. The effectiveness of AGV dispatching rules appears to be dependent on the system being analyzed.