ABSTRACT
Transport models are used to evaluate new infrastructure and public transport services, varied levels of demand, and new ideas for demand management. Exploring these proposals virtually is easier than implementation and testing in situ. However, existing models are based around traditional forms of transportation. As part of a feature analysis using a case study approach, three different simulation packages (a simple custom-developed package, traffic microsimulation, and agent-based simulation) are used to develop and demonstrate simulations of demand-responsive transportation (DRT) and analyze the advantages and disadvantages of each simulation approach for evaluating DRT. While the simulations display some relational replication (meaning they produce similar relational patterns with respect to certain variables), they do not show distributional replication (that is, the value of the results is not statistically similar), meaning that under- or over-estimation of predicted travel could occur. Recommendations for the application of each modeling approach are made.
Acknowledgements
The authors are grateful to John Haasz and John McDonald (Delphi development), Michal Maciejewski (MATSim+dvrp), Daniel Krajzewicz (SUMO), and Shubham Jain (preliminary investigations into SUMO).
Disclosure statement
No potential conflict of interest was reported by the authors.