The vehicle scheduling problem with intermittent customer demands

The vehicle scheduling problem (VSP) is a problem of interest to many researchers and practitioners. The general VSP is concerned with minimizing the total distance associated with a fixed set of vehicle routes by determining which vehicles will satisfy the demands at various customer locations. The design of most general distribution systems usually involves the assignment of permanent customer routes even though customer demand patterns are often stochastic in nature. The traditional routing designs implicitly assume that each customer will be visited each period for each route. In reality, a certain number of customers may not require a delivery for a particular period. In this research, we consider three alternative approaches for modelling the VSP with uncertain customer demands. Our solution methodology is a simple but practical heuristic which captures the best of all three approaches in order to handle the dynamic nature of vehicle scheduling under uncertain demand. An experimental design was also implemented to show the sensitivity of the proposed solution methodology to various realistic scheduling situations.