Worst case performance scheduling facing uncertain disruption in a continuous casting process

Several disruptions (e.g., machine failure, quality problem, etc.) in a steel production often occur in practice and lead to a delay in the production. Many schedule changes in a short period of time lead to an unstable production. Therefore, a robust predictive scheduling, which takes the disruption effect into account, is a more suitable choice for handling the daily small disruption. In this paper, a worst case performance scheduling via Minimax optimization for a multi-continuous casting is presented. By this approach, each uncertain disruption event is defined to be a possible scenario. A set of factory maintenance information (e.g., a machine reliability/availability) is utilized for constructing and illustrating the uncertain disruption model. Two loops of Genetic Algorithm (GA); one searching the worst scenario for each feasible schedule and another searching an optimal schedule, are performed. The objective function is formulated in terms of the steel production costs. The robust performance based on the factory data is demonstrated via Monte Carlo simulation. It shows that the worst case performance schedule can robustly handle the uncertain daily disruption.