Modelling ad-hoc DRT over many days: A preliminary study

Abstract

Traditional transport models focus on commuter traffic in private vehicles or public transport. However, new methods of travel are gaining traction, both in discussions and implementation: carsharing, shared taxis/buses, autonomous vehicles. Modelling these requires more attention on the interaction between people and place than traditional modes. As part of a larger project on mobility on demand, shared flexibly-scheduled vehicles (also known as demand-responsive transportation, DRT) have been explored using simulations. These schemes can be designed in different ways with respect to the resolution of stops and timings, and the potential patronage. In a pure ad-hoc scheme, there will be no timetable or route. These schemes are attractive in areas where mass transit cannot provide a reasonable level-of-service in a financially and environmentally sustainable manner, such as fringe areas and small towns, however could also find a niche providing access to mass transit in larger cities. One still-outstanding problem is predicting demand for DRT. Previous experiments have focused on trips, but have rarely looked into the choice to (continue to) use a service, especially when wait and travel times could change from day to day depending on the passengers using the service, and the effects on medium-term demand. Can an equilibrium point be found between supply and demand? If too many people use the service, wait and travel times increase too much for it to be a feasible option for some passengers, and so they take other modes or decide not to travel. However, the then decreased demand leads to good level-of-service for the remaining passengers; other customers might then return, which leads to worsening level-of-service, and so on. Another case occurs when an initially demand-responsive service starts to become “regular” in its routes and timings, in which case it is more efficient to create a fixed-route service. Using agent-based modelling, we explore how demand for DRT changes on a daily basis. Passengers make decisions about whether to use DRT to participate in an activity, using information and satisfaction levels from previous trips and publicly-available information about service levels. Results reported include the activity levels and locations of the population, as well as the performance of the transport network. Experimenting with a simplified environment shows that mismatched supply and demand lead to fluctuating performance of the DRT systems.