# A Novel Bin Design Problem and High Performance Algorithm for E-commerce Logistics System

## TL;DR

This paper introduces a new bin design problem for e-commerce logistics, aiming to minimize surface area by optimizing bin sizes, and proposes a high-performance heuristic algorithm that outperforms existing methods in efficiency and cost reduction.

## Contribution

The paper formulates a novel bin design problem with variable bin sizes and develops a dynamic programming-based heuristic algorithm, DPTS, that improves efficiency and cost savings over traditional methods.

## Key findings

- DPTS outperforms greedy local search by 5.8% in total cost.
- DPTS requires only 1/50 of the computational resources of GLS.
- Numerical experiments validate DPTS's efficiency and effectiveness.

## Abstract

Packing cost accounts for a large part of the e-commerce logistics cost. Mining the patterns of customer orders and designing suitable packing bins help to reduce operating cost. In the classical bin packing problem, a given set of cuboid-shaped items should be packed into bins with given and fixed-sizes (length, width and height) to minimize the number of bins that are used. However, a novel bin design problem is proposed in this paper. The decision variables are the geometric sizes of bins, and the objective is to minimize the total surface area. To solve the problem, a low computational-complexity, high-performance heuristic algorithm based on dynamic programming and depth-first tree search, named DPTS, is developed. Based on real historical data that are collected from logistics scenario, numerical experiments show that the DPTS out-performed 5.8% than the greedy local search (GLS) algorithm in the total cost. What's more, DPTS algorithm requires only about 1/50 times of the computational resources compared to the GLS algorithm. This demonstrates that DPTS algorithm is very efficient in bin design problem and can help logistics companies to make appropriate design.