936. Dynamic time-window updates in last-mile delivery: Follow Your Parcel
Invited abstract in session TD-56: Routing with Time Windows, stream Vehicle Routing and Logistics.
Tuesday, 14:30-16:00Room: Liberty 1.11
Authors (first author is the speaker)
| 1. | Roshan Mahes
|
| 2. | Alex Kuiper
|
| Business Analytics, University of Amsterdam |
Abstract
Accurate delivery time windows are crucial for customer satisfaction in last-mile logistics, influenced by factors such as route optimization and driver behavior. This work introduces novel analytics to study route performance, uncover route adherence, and explore their interdependencies, classifying four types of routes. These insights are used to provide a comprehensive methodology for generating time windows and time-window updates.
Real-life data shows that exponential distributions effectively model the time between consecutive deliveries, with means decreasing throughout the day. We approximate these using a regression model based on the position in the route and the route length. Next, each customer receives an individual expected delivery moment within a prediction interval. Besides the speed of generating these intervals, they outperform statically generated intervals of, say, 2 hours for nearly all route types. Moreover, we further leverage this approach by dynamically providing updates, which can be implemented as push or live updates.
Concluding experiments confirm the effectiveness of reducing time-window width during the route without significantly jeopardizing the performance of delivering within these intervals. Finally, to meet customer expectations for more precise delivery information, PostNL, the leading postal and parcel delivery company in the Netherlands, has implemented a customized version of the proposed methodology.
Keywords
- Logistics
- Analytics and Data Science
- Transportation
Status: accepted
Back to the list of papers