EURO 2025 Leeds
Abstract Submission

1799. Smarter demand allocation: Improving our VRP solver by redistributing demand between depots and delivery days

Invited abstract in session TB-56: Multi-Period Vehicle Routing Problems, stream Vehicle Routing and Logistics.

Tuesday, 10:30-12:00
Room: Liberty 1.11

Authors (first author is the speaker)

1. Michael Sutherland
Datasparq
2. Louisa Sober
Datasparq
3. Ying Tan
Datasparq
4. Jessica McQuade
Datasparq
5. Adil Rahman
Data Science, Datasparq

Abstract

In this presentation, we explore two key extensions to our ALNS-based vehicle routing optimiser, enhancing its flexibility while maintaining efficiency and scalability. These extensions—depot volume balancing and delivery day balancing—allow logistics networks to adapt dynamically to fluctuating demand.

The depot volume balancing extension addresses peak demand periods where depot capacity is exceeded. To manage this, we developed methods for reallocating customers between depots, ensuring a more balanced order distribution. While the primary goal was feasibility, empirical results showed cost reduction in certain depot pairs, suggesting even their well-established assignments can benefit from optimisation.

The delivery day balancing extension investigates whether adjusting customer delivery days within a weekly schedule can improve routing efficiency. Unlike depot balancing, this extension is embedded within the core ALNS framework, leveraging parallelisation to explore a vast solution space efficiently. Experimental results for this extension were promising, demonstrating cost savings across a variety of different depots.

These extensions improved both feasibility and cost-effectiveness, demonstrating how targeted optimisation enhances logistics performance. This talk will provide insights into the methodologies, challenges, and impact of these approaches, offering practical takeaways for large-scale routing problems.

Keywords

Status: accepted

Back to the list of papers