Optimization heuristics for supplies of assembly linest

Abstract

This paper deals with a heuristic approach to material supplies of assembly lines (e.g. automotive industry). A modern method for supplying assembly lines with material is using the so called 'milk run' - trains supplying not only one point in assembly production lines but several points. A graph model is used. An analytical solution for creation of trains is not known; most probably it does not exist. Solutions using "brute force" may be very slow. They cannot be used for more than a dozen demands. A repeated random selection of n-tuples of transport demands and building of trains from this selection could be a good way to solve this task. A model of assembly production lines has been developed and the speed of convergence of random selections to a suboptimal solution has been calculated and measured. A thousand selections give good results. These heuristic results have been compared with some deterministic strategies (nearest demand, building of n-tuples).