Abstract
In densely populated urban areas, increasing private vehicle usage continues to overwhelm existing transport infrastructure. Peer-to-peer ride-sharing—where drivers are also travelers—offers a promising solution by shifting mobility from ownership to usage. However, the real-world effects of ride-sharing on urban networks depend on dynamic travel behavior, network congestion, and operational constraints. We introduce a novel ride-sharing scheme that minimizes total social costs—including travel time, emissions, and fuel consumption—within a dynamic, multi-modal transport framework. The proposed method integrates a linear matching optimization with the agent-based simulator METROPOLIS2 to jointly model route, mode, and departure time choices under endogenous congestion. When applied to the Île-de-France region during the morning peak, the socially optimal matching scheme results in 4.0% of trips using ride-sharing, reducing vehicle kilometers by 5.8% and daily CO2 emissions by 342 tonnes. These results demonstrate the potential of ride-sharing to generate substantial environmental and efficiency gains, particularly when supported by targeted incentives.
Keywords
Dynamic congestion; Environment; Large-scale optimization; Ride-sharing; Socially optimal matchingBibtex (click to select)
@unpublished{GhoslyaEtAl2025,
author = {Ghoslya, Samarth and Javaudin, Lucas and de Palma, André and Delle Site, Paolo},
title = {{Ride-sharing, congestion, departure-time and mode choices: A social optimum perspective}},
year = 2025,
month = 09,
doi = {https://dx.doi.org/10.2139/ssrn.5465467},
}