87. On the Feasibility of Strategic Delivery Splitting: A Novel Approach to Improve Courier Availability in High-Demand Food Delivery Markets
Invited abstract in session WB-10: Collaborative delivery, stream Mobility, Transportation, and Traffic.
Wednesday, 10:45-12:15Room: H16
Authors (first author is the speaker)
| 1. | Ann-Kathrin Meyer
|
| Universität Münster | |
| 2. | Ekin Ugurel
|
| University of Washington | |
| 3. | Tobias Brandt
|
| Universität Münster |
Abstract
The rapid growth of online food delivery platforms has transformed food delivery from an occasional convenience to an essential aspect of urban food consumption. However, this expansion has introduced significant operational challenges, particularly during peak demand periods when courier availability becomes a major constraint. Traditional optimization strategies, such as batching orders, struggle to balance efficiency with courier availability. This study introduces strategic delivery splitting as a novel approach which divides deliveries into two segments—restaurant to transfer point and transfer point to customer—with different couriers handling each segment. By allowing couriers to remain closer to high-demand zones, this method reduces firstmile time and assignment delays, potentially improving overall system efficiency. Our research makes four key contributions. First, we develop an agent-based model based on real-world food delivery data from Meituan to simulate the delivery process. Second, we formalize the splitting decision as a partially observable Markov decision process and apply a Q-learning approach to derive an optimal policy. Third, we demonstrate that strategic splitting can reduce delivery delays by approximately 37% during courier shortages. Finally, we identify specific conditions under which this approach outperforms traditional direct delivery, providing actionable insights for food delivery platforms. Our findings have significant implications for risk management in food delivery operations. By optimizing courier allocation, strategic delivery splitting offers a robust solution for maintaining service quality during high-demand periods.
Keywords
- Simulation
- Machine Learning
- Strategic Planning and Management
Status: accepted
Back to the list of papers