| Abstract: | This paper examines a generalization of the vehicle-dispatch problem in which time constraints as well as stochastic demand and travel times are considered. The objective is to minimize total travel time subject to vehicle load capacities and sequencing restrictions, meaning that certain points may be serviced only in specified time intervals. Intra-city travel times are estimated via multiple regression, and an effective heuristic is tested on single-estimate travel times. A GPSS simulation is used to evaluate the heuristic on stochastic travel times and also to examine the effects of other decision variables such as route configuration. Computational experience with an actual refuse collection problem involving 163 points per day is reported. |
