A unified framework for vehicle routing problems with stochastic travel and service times

Vehicle routing problems (VRPs) are concerned with the design of efficient routes that deliver goods and services from (to) central depots to (from) customer locations, satisfying specific business constraints. In the last 50 years, a vast amount of research has been devoted to solve different VRP variants. Nonetheless, most of the solution methods for VRPs are based on the premise that problem parameters such as travel times and customer demands are known in advance. However, in a practical setting, more often than not the problem parameters are uncertain and neglecting their stochastic nature may result in poor routing decisions. In this research, we tackle a family of problems that have received relatively little attention in the literature: the vehicle routing problems with stochastic travel and service times. The main contributions of this research are twofold: proposing a unified framework based on queueing theory to model stochastic travel and service times and illustrating how this framework can be embedded into a routing optimization engine to solve different problem variants.