Production optimization and scheduling in a steel plant: Hot rolling mill

In this presentation we investigate the production scheduling of a hot rolling mill in a steel plant. The production schedule must adapt to strict production rules derived from metallurgic and physical constraints. Successful manual planning requires therefore in practice thorough process knowledge and years of operating experience. In our optimization-based approach, we first build parts of the rolling programs using intelligent heuristics and compose these parts to fully feasible programs by solving a min-cost-flow problem. In a second step, the built programs are scheduled using a mixed integer linear programming formulation in order to obtain an optimal schedule that violates as few order due dates as possible. Finally, we describe how to embed the current system within a standard IT environment, connecting the business level to the plant layer.