A heuristic algorithm to improving the coil slitting process in the steel industry

Abstract

The steel industry is constantly facing problems and challenges that require optimisation to improve the production process. We present an algorithm to address a major challenge, the slitting problem, for a specific Spanish company. This problem arises when large steel coils need to be cut into smaller strips. Given the highly heterogeneous stock (coils come from previous operations), selecting the most suitable coils and defining the cutting patterns become very complicated due to operational and customer constraints. The company aims to reduce the leftovers and increase the service level (the difference between the weight requested by the customer and the weight supplied). The algorithm is currently in production and was validated using the company’s data and compared with an exact model. Results significantly improved the company’s operations, achieving a 50% reduction in leftovers and a much better service level in minutes, as opposed to the hours the company previously required. Although there are Mixed Integer Linear Optimization models that provide an optimal solution in small cases, they are not a viable alternative for the company because they require excessive computational time (even, in some cases, to obtain feasible solutions) and use overly expensive commercial solvers.