Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Analysis of Transshipment in Three-Sided Meal Delivery Services via Microhubs

Linxuan Shi, Zhengtian Xu

TL;DR

This paper tackles intra-city meal delivery by introducing a microhub-based transshipment design that partitions the service region into $K$ pie-shaped sub-areas and batches orders at a central microhub. A continuous-approximation framework and a Jackson queuing-network model are developed to estimate two key metrics: vehicle miles traveled $Q$ and customer waiting time $W$, with TSP-based service times augmented by a radial term to reflect urban routing. The work benchmarks transshipment against a VRPPD baseline and validates the approach using a Meituan dataset, showing substantial VMT reductions and improved waiting times during high-demand periods, especially for larger service areas; results also indicate potential drawbacks in low-demand settings. Practical implications suggest deploying transshipment selectively during peak hours and larger-area service regions, with future work extending to multiple hubs, dynamic transitions, and endogenous market responses.

Abstract

This paper introduces and analyzes a novel transshipment strategy for meal delivery. In this approach, the service area is partitioned into smaller sub-areas, with deliverers assigned to operate exclusively within these sub-areas. Meanwhile, a centrally located microhub functions as a logistic depot to facilitate the batching and transfer of meal packages toward different sub-areas. We model the meal delivery system with transshipment using networked G/G/m queues and analytically approximate two critical system performance metrics -- customer waiting time and vehicle miles traveled -- to evaluate the effectiveness of the proposed strategy. The performance achieved by transshipment is benchmarked against that of the classic pickup-and-delivery strategy without transshipment, both predicted using continuous approximations. For the latter, we enhance the existing modeling by incorporating the delivery distance profiles of individual orders to better match the meal delivery context. Our comparisons indicate that meal delivery via transshipment outperforms the non-transshipping counterpart across both metrics under either high-demand or low-supply conditions, with particular advantages in servicing larger areas or handling long-distance orders. This conclusion is corroborated by a numerical experiment using empirical meal delivery data from Meituan, which suggests that an optimally configured transshipment strategy can significantly improve service performance for both customers and deliverers during peak lunch hours and in the busiest districts. While transshipment continues to reduce vehicle miles traveled by deliverers during non-peak hours, it results in longer customer waiting times compared to the benchmark without transshipment as demand decreases.

Analysis of Transshipment in Three-Sided Meal Delivery Services via Microhubs

TL;DR

This paper tackles intra-city meal delivery by introducing a microhub-based transshipment design that partitions the service region into pie-shaped sub-areas and batches orders at a central microhub. A continuous-approximation framework and a Jackson queuing-network model are developed to estimate two key metrics: vehicle miles traveled and customer waiting time , with TSP-based service times augmented by a radial term to reflect urban routing. The work benchmarks transshipment against a VRPPD baseline and validates the approach using a Meituan dataset, showing substantial VMT reductions and improved waiting times during high-demand periods, especially for larger service areas; results also indicate potential drawbacks in low-demand settings. Practical implications suggest deploying transshipment selectively during peak hours and larger-area service regions, with future work extending to multiple hubs, dynamic transitions, and endogenous market responses.

Abstract

This paper introduces and analyzes a novel transshipment strategy for meal delivery. In this approach, the service area is partitioned into smaller sub-areas, with deliverers assigned to operate exclusively within these sub-areas. Meanwhile, a centrally located microhub functions as a logistic depot to facilitate the batching and transfer of meal packages toward different sub-areas. We model the meal delivery system with transshipment using networked G/G/m queues and analytically approximate two critical system performance metrics -- customer waiting time and vehicle miles traveled -- to evaluate the effectiveness of the proposed strategy. The performance achieved by transshipment is benchmarked against that of the classic pickup-and-delivery strategy without transshipment, both predicted using continuous approximations. For the latter, we enhance the existing modeling by incorporating the delivery distance profiles of individual orders to better match the meal delivery context. Our comparisons indicate that meal delivery via transshipment outperforms the non-transshipping counterpart across both metrics under either high-demand or low-supply conditions, with particular advantages in servicing larger areas or handling long-distance orders. This conclusion is corroborated by a numerical experiment using empirical meal delivery data from Meituan, which suggests that an optimally configured transshipment strategy can significantly improve service performance for both customers and deliverers during peak lunch hours and in the busiest districts. While transshipment continues to reduce vehicle miles traveled by deliverers during non-peak hours, it results in longer customer waiting times compared to the benchmark without transshipment as demand decreases.
Paper Structure (26 sections, 3 theorems, 31 equations, 15 figures, 1 table)

This paper contains 26 sections, 3 theorems, 31 equations, 15 figures, 1 table.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Distribution system without transshipment
  4. Distribution system with transshipment
  5. Integrated Meal Delivery System with Transshipment
  6. Meal delivery with transshipment
  7. Lifetime of meal delivery orders
  8. Queuing network model
  9. Service time approximations
  10. System performance metrics
  11. A VRPPD Benchmark without Transshipment
  12. A heuristic for meal delivery without transshipment
  13. Service rate estimations
  14. System performance metrics
  15. Numerical Analysis
  16. ...and 11 more sections

Key Result

Proposition 1

For a circular sector of radius $R$ containing $N$ nodes, the expected farthest distance, $\mathop{{}\mathbb{E}}[R']$, is given by $\mathop{{}\mathbb{E}}[R'] = \frac{2N}{2N+1}R$, where $\mathop{{}\mathbb{E}}[R']$ is independent of the sector's angle.

Figures (15)

  • Figure 1: Comparison between transshipment in line-haul logistics and meal delivery.
  • Figure 2: Lifetime of individual meal delivery orders
  • Figure 3: Cross-zonal meal delivery process as a queuing network
  • Figure 4: Scenario enumeration of state transitions between the types of preceding and succeeding points visited consecutively by individual deliverers.
  • Figure 5: Comparison between measured and predicted mean and variance of the traveling distance in TSP tours w.r.t. 1) radius size $R$, 2) number of points $n$, and 3) number of divisions $K$.
  • ...and 10 more figures

Theorems & Definitions (3)

  • Proposition 1
  • Proposition 2
  • Proposition 3