Solving 2-Dimensional Multiple Length Cutting Stock Problem with MILP

Abstract

This paper considers a 2-dimensional cutting stock problem. Unlike previous works, we study multiple lengths in each cutting pattern and 3 types of production loss including trim loss, setup loss, and over-production loss. The problem is first formulated as a MINLP model to minimize total production loss, which has a non-convex feasible region. All the model constraints are developed based upon real settings. To solve the model, the MINLP model is transformed to a MILP model by prespecifying all cutting patterns. Compared to the manual planning method currently used by the company, the proposed method can provide 20.3% improvement. Other benefits include the saving of man hours in the planning process, and the planning process can be performed by any workers instead of an expert.