A Constraint Programming Approach to Electric Vehicle Routing with Time Windows

Abstract

The Electric Vehicle Routing Problem with Time Windows (EVRPTW) extends traditional vehicle routing to address the recent development of electric vehicles (EVs). In addition to traditional VRP problem components, the problem includes consideration of vehicle battery levels, limited vehicle range due to battery capacity, and the presence of vehicle recharging stations. The problem is related to others in emissions-conscious routing such as the Green Vehicle Routing Problem (GVRP). We propose the first constraint programming (CP) approaches for modeling and solving the EVRPTW and compare them to an existing mixed-integer linear program (MILP). Our initial CP model follows the alternative resource approach previously applied to routing problems, while our second CP model utilizes a single resource transformation. Experimental results on various objectives demonstrate the superiority of the single resource transformation over the alternative resource model, for all problem classes, and over MILP, for the majority of medium-to-large problem classes. We also present a hybrid MILP-CP approach that outperforms the other techniques for distance minimization problems over long scheduling horizons, a class that CP struggles with on its own.