How system configuration shapes demand: agent-based simulation of mobility-on-demand systems

To establish Mobility-on-Demand (MoD) and autonomous Mobility-on-Demand (AMoD) services as part of sustainable transport systems, a forward-looking configuration process is indispensable. Modern methods such as agent-based transport simulation are particularly well suited for this task, as they combine the representation of individual travel behavior with the physical interactions of the transport system. This thesis demonstrates how such a simulation framework can be applied as a decision-support tool in the context of a real-world pilot project. For this purpose, a new open-source model for the county of Kelheim, Germany, is presented, which is calibrated for the first time with real-world AMoD booking data. The key results underscore the need to carefully balance service area and fleet size. In the specific project context, the findings indicate that expanding the fleet requires a corresponding enlargement of the service area to maintain efficiency. These insights directly informed the decision-making process in the project. Apart from decision-support, this thesis contributes to the understanding of fundamental interactions between demand and supply in MoD systems. The findings suggest that configuration parameters determining the demand potential, such as service area or operating times, have a linear effect on realized demand. In contrast, parameters shaping service quality, such as fleet size or vehicle speed, are observed to have superlinear impacts on realized demand. This highlights the significant impact of service quality on the adoption of MoD, particularly in the early phases of deployment with limited fleet sizes. Furthermore, it is found that typical demand-anticipatory rebalancing strategies result in uneven service quality across the area. It is shown that in contrast to these demand-anticipatory rebalancing strategies, maintaining an equal vehicle density or vehicle-to-population ratio can improve the spatial balance of service quality. The findings imply that leaving MoD systems solely to profit-oriented operators would risk concentrating services in dense city centers while neglecting suburban and rural areas, underlining the need for regulatory integration into public transport (PT). Looking forward, future simulation research should address heterogeneity in user preferences, stronger intermodality with schedule-based PT services, and more advanced mode choice formulations. Beyond these aspects, it will also be important to investigate how the transition to true driverless operations (SAE Level 4) and the use of multifunctional vehicle concepts could shape the longer-term evolution of AMoD toward integrated Transport-as-a-Service models that combine passenger and freight transport.