Characterization of traffic dynamics in non-equilibrium ride-hailing mobility networks: A mesoscopic approach

Ride-hailing vehicles, private vehicles, and passengers are integral components of ride-hailing markets. Accurately characterizing the traffic dynamics driven by the spatio-temporal variations of these traffic flows is crucial for formulating operational strategies to realize sustainable ride-hailing services. From the mesoscopic perspective, we develop an integrated simulation model with high spatio-temporal resolutions. In a multi-class cell transmission model, we embed aggregate-ratio based decision-making mechanisms and bilateral matching between waiting passengers and idle vehicles in a large-scale non-equilibrium ride-hailing mobility network. At the individual level, the simulation model can capture the entire trip chain of passengers. Simultaneously, it can describe the cruising strategy of idle vehicles and the routing strategy of reserved/occupied/private vehicles. At the network level, it can depict the real-time space distribution of these multi-class traffic flows in the ride-hailing mobility network. We use empirical data, including road network density data, ride-hailing order, and trajectory data, to calibrate and verify the proposed simulation model. Sensitivity analyses based on simulation experiments indicate that the matching strategy, fleet size, and background traffic have diverse and significant effects on the operation performance of ride-hailing services.