Multi-objective optimization of demand responsive transit operations based on dynamic passenger requests using maximum time delay rate

Demand-responsive transit (DRT) offers on-demand service for comfortable and convenient trips. Despite these advantages, efficient DRT operation requires addressing several considerations. This study resolves the conflict between passengers wanting quick travel and operators seeking maximum revenue by formulating a multiobjective mixed-integer nonlinear programming model (MINLP) to maximize revenue and minimize total travel time. Additionally, DRT operators should balance the benefits of accepted passengers, concerned about increased travel time from new passengers, and requesting passengers who intend to use DRT. To address this, unlike previous studies with fixed time windows, this study introduces the maximum time delay rate (MTR), setting a proportional threshold for each accepted passenger's travel time based on their scheduled travel time, incorporating behavioral economics principles. In this view, the utility of increased or decreased time varies according to the scheduled travel time, considered a sunk cost. When the increased travel time from a new request is within the allowable range, the request is accepted, then the passenger decides whether to choose DRT over other modes. We apply our methodology to dy namic passenger requests generated from taxi data in Incheon, South Korea. For each combination of operational parameters of DRT, we plot a Pareto optimal set of revenue and total travel time. The results demonstrate the substantial influence of MTR and minimum fare distance on passenger numbers and travel time in DRT operations. This study's methodology and results help DRT operators and the public find desirable operation strategies.