Integrating Aggregate and Request-Based Demand Data in Line-Based Semi-Flexible Public Transport

• Applicant: Professorin Dr. Marie Schmidt
• Subject Area: Operations Management and Computer Science for Business Administration
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 566982931

Passenger transport on roads alone accounts for about 45% of the transport-related greenhouse gas emissions worldwide. Public transport contributes to emission reduction through the pooling of passenger journeys. By transporting passengers together, in a minibus, bus, tram, or train, the emissions per traveled kilometer are significantly reduced compared to a situation where each traveler travels individually by motorized transport. What is more, a well-functioning public transport system provides competitive, reliable, and inclusive mobility options, both in urban and in rural areas, to meet the mobility needs of the whole population and enhance their ability to participate in society. Public transport is particularly efficient in high-demand settings, but struggles when there is low demand. When offering frequent services under low demand, vehicles may run empty on part of their trajectories, leading to higher total emissions than if everyone had traveled individually by car. Where service frequencies are reduced to achieve higher pooling rates, the quality of service may deteriorate to an extent that makes public transport unattractive for a high proportion of travelers, resulting in further reduction of ridership. This difference in efficiency leads to a high inequality in mobility options. In this project, we want to rethink public transport services in medium and low-demand areas and suggest new approaches based on mathematical optimization. There are two co-existing paradigms for the temporal organization of public transport services: Regular public transport services operate based on predefined schedules or predefined headways, while on-demand services are tailored to a set of individual passenger requests, which are different each day. Models used in the literature for optimizing on-demand services differ from models for regular services: in scope, modeling of passenger journeys, capacity, and the determination of performance indicators. We argue that many of these differences are rooted in the different demand data used: regular services are planned based on aggregate demand data, while on-demand services are planned based on a set of individual passenger requests. In this project, we investigate how to optimally schedule services under both paradigms, and attempt to design an approach that combines the best features of both worlds. To do that, we concentrate on a particular spatial setting, namely a bus line that is operated in a semi-flexible way: not all stops need to be visited in each service, and a vehicle may turn before reaching the end of the line. We start by assuming full information and model the corresponding optimization problems as integer linear programs, and develop appropriate approaches to solve them. In later workpackages, we develop approaches to use both aggregate demand data and individual requests for the dynamic version of the problem.

• DFG Programme: Research Grants