Dynamic evolutionary multiobjective optimization for open-order coil allocation in the steel industry

In terms of improving coil resource utilization and customer satisfaction, most steel companies are still far behind in these quality objectives and need significant improvements. Scientific approaches to production and operations planning are critical to improve its situation. Motivated by this phenomenon, this study investigates a challenging operations decision problem on allocating open-order coils to customer orders in the steel industry. In this article, a dynamic multiobjective optimization model is formulated to optimize the total mismatching costs, surplus inventory and coil utilization considering various practical dynamisms, such as production changes and unscheduled arrivals of new customer orders. To address this problem, we propose a multiobjective evolutionary algorithm based on knee-driven change response strategy and Pareto local search mechanism, so called KPLSEA. This proposed approach, which combines the information of decision-making stages with the evolutionary search, significantly reduces the computational cost and promotes the evolutionary search to converge quickly. Extensive experiments are performed on real-world production benchmark instances. Computational comparisons with other state-of-art algorithms validate that the proposed algorithm could generate effective and practical solutions in dynamic environments as well as provide the decision makers with a satisfied near-optimal solution.