A stochastic scheduling problem issued from iron and steel industry : Some results

Abstract

A stochastic scheduling problem with resource constraints arising in iron and steel industry is discussed. It concerns the utilization of highly flexible automated and integrated bricklaying systems inside converters located on different sites. The stochastic aspect concerns the time windows to which one task of bricklaying has to be assigned. The resource constraint comes from the limited number of bricklaying systems. For this problem simulation is used first in order to compare some strategies which can be used for the on-line decision problem. We then try to analyse them : was the chosen number of bricklaying systems adequate ? To answer this question two ways are proposed : the search of a priori lower and upper bounds under particular assumptions and the study of the deterministic problem a posteriori after a simulated execution. The first section describes the considered problem. The second one presents the lower and upper bounds. The third section presents very briefly the simulation strategies for the on-line decisions. The fourth one gives the general approach for an "a posteriori" analysis and the fifth section presents a numerical sample.