Analyzing the impact of travel time estimations for large-scale mobility-on-demand systems in stochastic urban networks

Mobility-on-Demand (MoD) services could offer a promising solution to increasing congestion and spatial pressures in urban areas. However, their reliability and operational efficiency in stochastic urban networks critically depend on accurate travel time estimates. This study assesses the network-wide and operational impacts of MoD systems in stochastic traffic flow under varying demand, and examines how different travel time estimation methods and data sources affect the efficiency and reliability of the system. A coupled framework is presented that integrates a MoD model with microscopic traffic simulation, capturing information flow between the network, fleet operator, and both private and MoD vehicles. The study examined a MoD system within a citywide network replacing private trips. Results show that discrepancies between estimated and actual travel times affect performance, emphasizing the value of data-driven estimations. Real-time data improves outcomes in high-demand conditions, historical data suits low-demand scenarios, and a hybrid approach is effective across all conditions.