Simulating on-demand ride services in a Manhattan-like urban network considering traffic dynamics

Abstract

With the rise of the mobile internet economy, on-demand ride services (or ride-sourcing services) provided by transportation network companies (TNCs) have become prosperous as an emerging mobility service mode. In this paper, we present a simulation framework to study how such an on-demand ride services platform should optimize pricing strategies considering traffic dynamics, that it, the platform charges the passengers and pays the ride-sourcing drivers with the wage. Specifically, we use the cellular automaton model to simulate route choice behavior of both regular cars that impact background traffic dynamics and ride-sourcing cars that provide on-demand ride services in the Manhattan-like urban network. In this model, passengers are sensitive to the waiting time and price, while ride-sourcing drivers are sensitive to the wage. The proposed model framework is capable of simulating the on-demand travel behavior of multiple types of stakeholders, including the on-demand ride services platform, ride-sourcing drivers and cars, and passengers. The model also realizes the process of trip requests, passenger–driver matching, dispatching, and dynamic path planning. We quantitatively evaluate the impacts of various operational strategies on ride-sourcing and passenger behavior. The results show that it is optimal for the platform to charge a higher price and offer a higher payout ratio as demand increases, while the platform should lower its payout ratio as the number of service providers increases. Furthermore, the optimal price and payout ratio are not necessarily monotonic when the travel distance increases. This paper is one of the first quantitative studies that simulate on-demand ride services in a microscopic traffic environment with mixed regular vehicles and ride-sourcing vehicles.