A dynamic day-ahead paratransit planning problem

We consider a dynamic planning problem for the transport of elderlyand disabled people. The focus is on a decision to make one day ahead:which requests to serve with own vehicles, and which ones to assign tosubcontractors, under uncertainty of late requests which are gradually revealedduring the day of operation. We call this problem the DynamicDay-ahead Paratransit Planning problem. The developed model is a nonstandardtwo-stage recourse model in which ideas from stochastic programmingand online optimization are combined: in the first stage clusteredrequests are assigned to vehicles, and in the dynamic second-stageproblem an event-driven approach is used to cluster the late requests oncethey are revealed and subsequently assign them to vehicles. A genetic algorithmis used to solve the model. Computational results are presentedfor randomly generated data sets. Furthermore, a comparison is made toa similar problem we studied earlier in which the simplifying but unrealisticassumption has been made that all late requests are revealed at thebeginning of the day of operation.