Problem instances for the two-dimensional bin packing problem with multiple levels of prioritization

This repository contains a collection of 90 problem instances for the two-dimensional bin packing problem with multiple levels of prioritization (2D-BPP-P). These instances were generated to support research that integrates spatial optimization with traditional bin packing, a scenario where the arrangement of items is as critical as the efficiency of the packing itself.

Dataset Structure and Contents:

The dataset comprises 90 unique, computationally generated problem instances. Each instance defines a set of rectangular items to be packed into a single, larger rectangular bin with a designated access point.

The data for each instance includes:

Bin Dimensions: The length (L) and width (W) of the single bin. The bin length is intentionally extended to provide sufficient space for prioritization-based layouts, moving beyond simple space minimization.

Item characteristics: For each rectangular item, the dataset specifies its length, width, and group affiliation. Items may be rotated by 90 degrees. Item dimensions are generated to reflect realistic aspect ratios (1:1 to 1:3), analogous to military vehicles or varied package sizes.

Group Structure: Items are organized into 1 to 5 groups, with group sizes categorized as small (1–4 items), medium (5–8), or large (9–12). Group item dimensions are classified as homogeneous, weakly heterogeneous, or strongly heterogeneous.

Prioritization Scheme: Each item is assigned a group-level priority and an item-level priority within its group. This two-tiered system allows for the modeling of complex operational requirements, such as maintaining cohesion among functionally related items while ensuring high-priority items are positioned near a bin access point. The raw priority data is provided, from which a full prioritization matrix can be constructed to weigh the objective function in optimization models.

Reuse Potential:

This dataset is intended for researchers and practitioners in operations research, industrial engineering, computer science, logistics, and other fields. The instances are useful for benchmarking and developing new solution methodologies for combinatorial optimization problems that blend packing and facility layout concepts. Potential applications include:

Validating and comparing the performance of exact algorithms, heuristics, and metaheuristics for spatial optimization problems.
Studying the trade-offs between space utilization and operational priorities in logistics applications such as military combat loading.
Extending the problem to multi-bin scenarios, or incorporating additional real-world constraints such as load balancing or non-adjacency requirements.

Legal and Ethical Considerations:

The data is synthetically generated and does not contain any sensitive, confidential, or proprietary information. There are no legal or ethical restrictions on its use. The authors encourage its reuse and dissemination for academic and research purposes, with appropriate citation to the associated article(s).

Methods
Instance generation is explained in detail in the following paper (see Section 5.1):

Kirschenman, W.K., Heese, H.S., Kay, M.G., King, R.E., McConnell, B.M., 2025. The 2-D bin packing problem with multiple levels of prioritization: A spatial optimization perspective. Working paper. https://www.lib.ncsu.edu/resolver/1840.20/44301.