Performance Metamodels for Dial-a-Ride Services with Time Constraints

Explicit performance models of a transit system are often very useful for system design, optimization, alternative comparison, and gaining insights into relevant system relations. In this paper, three performance metamodels have been developed for many-to-many dial-a-ride service, in which flexible routes and schedules are provided and service quality is guaranteed by time constraints. The models predict minimum vehicle fleet size requirements, the average deviations from service times desired by passengers and the average passenger ride time ratios. A simulation-based response surface methodology is used to model the functional relations between performance and contributing factors through experiments and statistical analysis. A detailed vehicle routing and scheduling algorithm and passenger time constraints, which are oversimplified or omitted in most other analytical approaches, are incorporated in the simulation experiments from whose results our models are statistically estimated. A face-centered central composite design is used to determine the experiment design points. The metamodels are validated using an additional set of randomly generated data. The resulting models are relatively simple in structure, inexpensive to use and fairly robust.