Jostle heuristics for the 2D-irregular shapes bin packing problems with free rotation

Abstract

The paper investigates the two-dimensional irregular packing problem with multiple homogeneous bins (2DIBPP). The literature on irregular shaped packing problems is dominated by the single stock sheet strip packing problem. However, in reality manufacturers are cutting orders over multi-stock sheets. Despite its greater relevance, there are only a few papers that tackle this problem in the literature. A multi-stock sheet problem has two decision components; the allocation of pieces to stock sheets and the layout design for each stock sheet. In this paper, we propose a heuristic method that addresses both the allocation and placement problems together based on the Jostle algorithm. Jostle was first applied to strip packing. In order to apply Jostle to the bin packing problem we modify the placement heuristic. In addition we improve the search capability by introducing a diversification mechanism into the approach. Furthermore, the paper presents alternative strategies for handling rotation of pieces, which includes a restricted set of angles and unrestricted rotation. Very few authors permit unrestricted rotation of pieces, despite this being a feature of many problems where the material is homogeneous. Finally, we investigate alternative placement criteria and show that the most commonly applied bottom left criteria does not perform as well as other options. The paper evaluates performance of each algorithm using different sets of instances considering convex and non-convex shapes. Findings of this study reveal that the proposed algorithms can be applied to different variants of the problem and generate significantly better results.