Oliver Herr

• The tasks of steel production planning and control have a major impact on the logistics target achievement and therefore on the competitiveness of the company. The planning process is known to be extremely difficult, with various incompatible local constraints at the different production stages. As a consequence, only limited amount of constraints can be respected in higher planning levels. Detailed production planning at the different production stages has the task to derive production programs that are able to respect all local constraints and at the same time lead to appropriate target achievement. Current approaches developed for detailed continuous casting planning are not able to quickly provide alternative solutions and therewith enable decision support for conflicting objectives. Within this thesis, the detailed continuous casting problem is presented in detail. A new approach to decompose the problem is described. With this decomposition, the problem can be treated as a single machine scheduling problem and effective meta heuristics developed for similar problems can be exploited. Further, with the chosen decomposition it is possible to respect the consumption of hot metal within the scheduling of charges. This important practical constraint could not be respected within the continuous casting problem in the past. Besides the hot metal consumption, setup families and maximum batch sizes are considered in the scheduling model. MILP models are presented for the different extensions of the basic scheduling model. A iterated local search procedure is presented and the effective is shown based on the comparison with a commercial solver. The findings obtained from the scheduling research is transferred into a decision support system for the detailed continuous casting planning. Based on an industry case study, the application of the developed tool is presented on a real industry situation.