A Dynamic Programming Model for Determining Continuous-Caster Configurations

Abstract

In the steel industry the continuous casting machine, or caster, can be used to eliminate a number of processing steps associated with the traditional steel production sequence from ingot through blooms to finished product. A given continuous caster can produce only a small number of bloom thicknesses, which creates a problem for selecting those continuous-caster configurations which would maximize caster utilization. A dynamic programming model was developed to assist Bethlehem Steel personnel to determine that set of caster configurations which would maximize the cast bloom tonnage that could be processed through one of the finishing mills. Without the aid of such a model, selecting the highest productivity options presents a difficult exercise because of two conflicting considerations: (1) as the number of caster-produced bloom thicknesses increases, the caster setup time and configuration complexity increase; and (2) as the number of thicknesses decreases, less cast tonnage can be processed through the finishing mill because of reheat-furnace and cooling-bed limitations. The model results were transmitted to plant management and are being used in conjunction with other information to determine the most economic configuration.