Efficient cutting stock optimization strategies for the steel industry.

Abstract

This study addresses a cutting stock problem in steel cutting industry by developing a mathematical model in which machine specifications and cutting conditions are constraints. The solution process involves three key steps: (i) Problem representation, where feasible cutting solutions are modeled based on pre-cut steel bars and customer orders, (ii) Problem space reduction, which reduces the problem space by eliminating suboptimal solutions and following manufacturer loss limits, and (iii) Optimal solution search, whereas the optimal solution is identified using a new Adaptive Pathfinding Optimization Algorithm. This algorithm combines a newly proposed Wandering Ant Colony Optimization with a brute force method, and uses specific conditions to determine which of these two approaches to be used to obtain the solution. The proposed algorithm can also be applied to other cutting stock problems, such as paper roll cutting, metal rod cutting, and wood plank cutting. The algorithm was applied to real customer orders in a steel manufacturer and showed significant benefits by reducing the number of planners from four to merely one person and decreasing the cutting planning time from six hours to under one hour. Additionally, the algorithm yields an average cost saving of USD 3.95 per ton, or 52.18% of the baseline.