Electric Vehicle Demand-Responsive Transport with Transit Integration

Integrated Demand-Responsive Transport (IDRT), which combines demand-responsive transport services with regular transit systems, is widely recognized as an effective strategy to mitigate the impact of standalone demand-responsive services on traffic congestion and the environment. However, successful implementation of such service must address customer inconvenience, as transfers between demand-responsive vehicles and transit services often discourage ridership. As the transition towards sustainable mobility accelerates, it is essential to incorporate electric vehicles into IDRT to enhance environmental benefits. Involving electric vehicles in the IDRT system also brings additional challenges, particularly in managing charging operations and ensuring service reliability. This dissertation introduces an Electric Integrated Demand-responsive Transport (EIDRT) service, in which electric buses operates with fixed-route transit to effectively meet customer demand.
To address the complexity of the EIDRT problem, we first investigate a meeting-point-based first mile feeder service utilizing electric buses. The objective is to minimize bus operational costs and customer inconvenience, including reducing customers’ waiting time at transit stations. Bus charging operations consider capacitated charging stations. A Mixed-Integer Linear Programming (MILP) formulation is developed using a layered graph structure and a metaheuristic solution algorithm is proposed. Computational experiments demonstrate that the metaheuristic produces good-quality solutions with around 1 or 2 minutes to solve 100-customer test instances. Results also show that the layered-graph significantly reduces computational time.
Next, we extend the meeting-point-based first-mile feeder service to a many-to-many EIDRT service that connects customers from their origins to destinations. A MILP formulation is proposed with the objective function minimizing both bus operational costs and customer travel times. To reduce customer inconvenience, the service incorporates synchronization between demand-responsive buses and transit departures, along with a maximum inter-modal transfer time. Capacitated charging stations are also included to reflect realistic recharging operations. The MILP formulation is built on a departure-expanded transit graph, incorporating the layered graph concept to better represent the transit network within a service area. A hybrid large neighborhood search algorithm is proposed to efficiently solve the problem, addressing the challenges of multi-modal routing and capacitated charging stations. The algorithm is benchmarked against eight-hour solutions by a MILP solver, demonstrating efficiency and better solution quality for up to 100 customers with two transit lines.
Lastly, the performance of the EIDRT is assessed by a set of experiments on scenarios reflecting real-world problem size and a case study. The findings provide valuable insights for operators regarding the trade-offs between operational costs and customer convenience, particularly focus on fleet size, buses’ state-of-charge, and transit networks. Results indicate that while the EIDRT service saves vehicle kilometers traveled compared with non-integrated demand-responsive services, maintaining a high-quality service might still require the same or a larger bus fleet size. Moreover, the results show that the service performs the best when the transit network in the service area is well-connected. The case study further compares the EIDRT service with existing transportation options, confirming significant reductions in bus operating costs relative to non-integrated demand-responsive services. Similar results are found that bus travelling costs are saved significantly compared to non-integrated demand-responsive services. In terms of customer travel experiences, EIDRT achieves substantially lower customer travel times compared to traditional public transport.