This document proposes a peer-to-peer grocery delivery system that matches customers who order groceries online (clients) with customers who are already going to the grocery store to shop (carriers). It analyzes different scenarios by varying the number of grocery stores and ratio of clients to carriers. The results show that ideal conditions occur when the number of carriers and clients are balanced. Across scenarios, the proposed matching system reduces total travel time compared to traditional truck delivery, especially at higher store counts and balanced client/carrier ratios.

1. Amin Sazavar
amin.sazavar@mail.utoronto.ca
A Peer-to-Peer Matching System for Grocery Home Delivery
Matthew Roorda
roordam@ecf.utoronto.ca
Marianne Hatzopoulou
marianne.hatzopoulou@mcgill.ca
• Collaborative consumption has made its way to our lives.
• Peer-to-peer sharing is a recent trend that is gaining popularity.
• Grocery companies could take advantage of peer-to-peer grocery
delivery to reduce the costs
• Groceries are the most universal commodities.
• Current grocery shopping methods:
• Regular grocery shopping
• Truck delivery (e-grocery)
Introduction
Overview of the proposed delivery system:
• Client : Customer who orders grocery online.
• Carrier : Customer who shops at a grocery store.
Proposed delivery scenarios:
• Scenario 1:
• Scenario 2:
• Scenario 3:
• Scenario 4:
• Analysis of the impact of changes in the number of stores
Results
Methodology
Methodology
• Comparison of total travel time across scenarios (4stores)
• Delivery truck travel times in scenario 2 and 4 (4stores)
Results
• The analysis is based on the Sioux Falls network which is composed of
24 zones and 76 links.
• Customers and stores are scattered randomly in the network zones.
• A (24 × 24) origin-destination matrix represents the number of trips
initiated and ended in the network.
• A (24 × 24) matrix represents the travel time between zones.
• A constant travel time of 2 minutes is assumed for all intra-zonal trips.
• Scenarios are simulated under six different client/carrier ratios as
follows:
• Scenarios are simulated in network with 4 different total number of
stores: 4,8,12 and 16.
Road network and agent locations
• Ideal conditions for matching occur when the number of carriers and customers is balanced
• on the basis of travel time, the matching system seems to be a promising method of shopping delivery, with potential reductions in fleet size for truck
delivery
Discussion and Conclusion
Carrier
Carrier
Client
Matched Client
Un-matched Client
Matched Client
CarrierClient
Client/carrier
ratio
0.25 0.5 1.0 1.5 2 4
No. of clients 300 500 750 900 1000 1200
No. of carriers 1200 700 750 600 500 300
Truck
Customer
Customer