Heuristic, storage space minimization methods for facility layouts served by looped AGV systems

An analytical method for predicting work-in-process storage requirements resulting from a fixed number of looping, automated guided vehicles (AGVs) serving a line layout is described. The method is based on a model of storage queue dynamics that predicts material flow rates and vehicle response times resulting from vehicle dispatching within a single loop system. The method is applied in the development of two heuristic, random number driven procedures designed to perform sequential search for WIP storage minimizing line layouts over alternative AGV fleet sizes. The first algorithm is a greedy, ‘CRAFT like’ procedure based on local improvement in storage requirements using random, pairwise exchanges of workcentre locations. The second algorithm is a modification to the greedy procedure using simulated annealing methods designed to avoid trapping at local minima. Limited computational studies using the methods are reported.