|
Title
|
|
|
|
Scattered storage assignment optimization : towards more efficient order fulfillment in e-commerce warehouses
| |
|
Author
|
|
|
|
| |
|
Abstract
|
|
|
|
| Various retail and e-commerce companies face the challenge of picking a large number of time-sensitive customer orders that include both a small number of items and multiple order lines. To reduce the unproductive work time of order pickers, several storage assignment policies have been proposed in the literature and in practice. In case of scattered storage assignment (SSA), items of the same type are intentionally distributed to multiple positions in the picking area to increase the likelihood that items belonging to the same order can be picked at nearby positions. We propose a scattered storage policy that, when determining the position where each individual item should be stored, tries to minimize the sum of pairwise distances (SPD) between all items belonging to the same order, including a drop-off point. A mixed integer programming (MIP) model is proposed to solve this scattered storage assignment (SSA-SPD) problem. We show that the solutions of the MIP solver can be improved by relaxing some variables and adding valid inequalities. We prove that the SPD objective for our scattered storage policy is 65% lower than for a traditional volume-based storage policy. Since solving the scattered storage assignment problem for large instances is challenging, we next develop an efficient heuristic based on a variable neighborhood search (VNS) metaheuristic to solve large instances in a reasonable computation time. The results show that VNS outperforms the exact approach with better solutions and runtimes. We then test our SSA-SPD policy by implementing a picking algorithm where items are picked up on an order-by-order basis. We develop a MIP model and a heuristic based on an adaptive large neighborhood search (ALNS) meta-heuristic for solving the picker routing problem. Our results show that the SSA-SPD policy helps reduce picking distances by up to 37% compared to a random scattered policy and by up to 57% compared to a volume-based policy. Finally, we test our scattered storage policy under the scenario where exceptionally large order lines may arrive at an e-commerce warehouse. So the question is whether scattered storage still makes sense when it comes to meeting demand for an atypical order consisting of large order lines. The results for picking routing distance under this storage policy show that up to 80% of large order lines SSA-SPD is still a good storage policy compared with a traditional volume based policy. |
| |
|
Language
|
|
|
|
English
| |
|
Publication
|
|
|
|
Antwerp
:
University of Antwerp
,
2023
| |
|
ISBN
|
|
|
|
978-90-5728-784-8
| |
|
Volume/pages
|
|
|
|
xvi, 96 p.
| |
|
Note
|
|
|
|
:
Cornelissens, Trijntje [Supervisor]
:
Sörensen, Kenneth [Supervisor]
| |
|
Full text (open access)
|
|
|
|
| |
|