# Joint Order Selection, Allocation, Batching and Picking for Large Scale Warehouses

**Authors:** Imran Khan, Olaf Maurer, Julius Pätzold, Paweł Pszona, Jan-David Salchow  
**arXiv:** [2401.04563](https://arxiv.org/abs/2401.04563)  
**Date:** 2024-01-09  
**Summary page:** https://www.summarizepaper.com/en/paper/2401.04563v1/

## Key points

- Order picking in picker-to-parts warehouses is a costly activity that requires efficient algorithms for optimization.
- Existing research has proposed various algorithms for order picking optimization, but many are not suitable for large-scale warehouses like those operated by Zalando.
- The researchers introduce a novel batching problem formulation tailored for mixed-shelves, large scale warehouses with zoning, incorporating order selection to enhance realism and optimization potential.
- Comparative analysis shows that even a basic greedy algorithm requires substantial runtime to solve real-world instances, but including order selection in the problem formulation improves solution quality by 40%.
- Practical insights from Zalando's operations reveal that batching algorithms prioritize creating batches that provide work for warehouse workers over specific orders, deviating from traditional assumptions.
- Flexibility in order selection processes is crucial for achieving more efficient and cost-effective solutions tailored to real-world logistics challenges in large-scale warehouse operations.

## Summary

In their study titled "Joint Order Selection, Allocation, Batching and Picking for Large Scale Warehouses," Imran Khan, Olaf Maurer, Julius Pätzold, Paweł Pszona, and Jan-David Salchow address the challenge of order picking in picker-to-parts warehouses. They highlight the significant cost associated with this activity and emphasize the need for efficient algorithms to optimize the process. Existing research has proposed various algorithms for order picking optimization; however, many are not suitable for large-scale warehouses like those operated by Zalando. Drawing on their experience with Zalando's batching system, the researchers introduce a novel batching problem formulation tailored for mixed-shelves, large scale warehouses with zoning. This formulation incorporates the selection of orders to be batched into the problem scope to enhance realism and optimization potential. The researchers present two baseline algorithms and conduct a comparative analysis using generated instances. The results reveal that even a basic greedy algorithm requires substantial runtime to solve real-world instances. Furthermore, including order selection in the problem formulation demonstrates significant potential for improving solution quality. The researchers compare average picklist costs per item across modified instances and observe a 40% enhancement in solution quality compared to previously defined instances. Moreover, they discuss practical insights from Zalando's operations where batching algorithms prioritize creating batches that provide work for warehouse workers over specific orders. This approach deviates from traditional assumptions that all available orders must be batched immediately. The computational results combined with real-world insights underscore the importance of flexibility in order selection processes. Overall, the study underscores the significance of considering order selection alongside batching optimization in large-scale warehouse operations. By incorporating these elements into algorithm design, researchers can achieve more efficient and cost-effective solutions tailored to real-world logistics challenges.