Plan Now, Ride Later: Simulation-Based Fleet Optimisation and Inclusive Service Design for Cities

Public transport systems of all sizes face a common dilemma in the mobility transition: introducing flexible, on-demand services to reduce private car dependency without incurring the financial risks of oversized fleets. While Mobility-on-Demand (MoD) promises efficiency, authorities often lack empirical data to determine the exact "tipping point" where service quality justifies the cost. The integration of MoD solutions is often hard to predict due to different city networks as well as supporting measures for the integration of MoD services. Furthermore, ensuring equitable access for vulnerable groups remains a critical challenge. This paper presents a novel simulation framework to bridge a multidimensional planning gap. Based on the funded project FLIPFLOP (Flexible Line and On-demand Public transport) the simulation focuses on a representative real-world city network in Austria. Using an open-source agent-based python simulation dynamic transit operations have been modelled by applying the Alonso-Mora algorithm for on-demand high- capacity ride-pooling. The approach supports the theory that the generation of on-demand ride pooling solutions depend on the relevant public city network, population size, job availability, cultural locations and many more variables. The research further focuses on a quantitative sensitivity analysis of fleet sizes under peak-hour stress tests, distinguishing itself by embedding diversity-specific constraints – such as wheelchair accessibility and varying digital literacy levels – directly into the dispatching logic. While project FlipFlop has focused on the city of Klagenfurt the wider simulation framework will in addition support the cities of Graz, Innsbruck, Linz and Vienna, where on-demand ride pooling can be integrated within public transport. Simulations reveal a non-linear efficiency frontier and have identified a distinct operational "sweet spot" at a specific fleet density that sustains a good service rate with average wait times stabilizing at a few minutes. Reducing fleet capacity by just a few percent below this threshold causes rejection rates to spike disproportionately, rendering the service unreliable. Preliminary simulations suggest accessibility features can be operationally integrated without evident service collapse, contingent on validation with real demand patterns and target user populations. Validating parameters through agent-based simulation facilitates providing a “Digital Twin” methodology that empowers planners to test service levels before procurement, offering modern mobility services that are tailored to user needs and cost-efficient. Suburban areas may be integrated seamlessly in the deployment design, as may be non- technical aspects, like compliance with the Austrian Transport Law, EU passenger rights, and equality laws. The results offer a validated roadmap for implementing inclusive on-demand transit that complies with modern service standards, ensuring that future investments are based on hard, dependable data rather than rough estimates.