Can dynamic ride-sharing reduce traffic congestion?

Can dynamic ride-sharing reduce traffic congestion? In this paper we show that the answer is yes if trip density high, which usually case in large-scale networks but not medium-scale where opportunities for sharing time and space become rather limited. When demand system can significantly improve conditions, especially during peak hours. Sharing compensate extra travel distances related to operating a mobility service. The situation entirely different small cities when shareability small, even fully optimized based on perfect prediction near future. reason simple, services increase total distance, simply means of combating trend without eliminating it high enough. This proposes complete framework represent functioning multiple steps tackle curse dimensionality solving problem. We address problem two city scales order compare densities. A scale 25 km2 with market 11,235 shareable trips network 80 205,308 service vehicles over 4-hour period rolling horizon 20 minutes. solutions are assessed using trip-based macroscopic simulation account congestion effect times may influence optimal solution obtained predicted times. outperforms most previous studies fleet management consider constant deterministic functions.