A three-stage solution for solving the cold rolling mill scheduling problem: an industrial case study

In steel manufacturing, the cold rolling mill (CRM) plays a pivotal role as the initial stage of the cold rolling process, governing thickness reduction and resource allocation for downstream operations. This paper addresses the industrial CRM scheduling problem with distinct narrow-to-wide (N2W) and wide-to-narrow (W2N) stages. We first establish a tailored mixed-integer linear programming model that considers practical constraints and stage-eligible coils. A dedicated three-stage solution framework is proposed to tackle this problem efficiently, in which the W2N subproblem is converted into an elementary shortest path problem with resource constraints by adopting the ϵ-constraint method, and an improved pulse algorithm integrated with multiple acceleration strategies is developed to solve the subproblem optimally.Numerical experiments on real industrial data show that our approach obtains high-quality solutions with computational efficiency and solution quality validated against Gurobi's results, confirming its practical effectiveness.