STOCHASTIC ELECTRIC VEHICLE ROUTING PROBLEM WITH THREE-DIMENSIONAL PACKING

This work addresses an extension of the Vehicle Routing Problem with Three-Dimensional Loading, in which the fleet is composed of electric vehicles and travel times are affected by time-dependent stochastic speeds. In this context, route feasibility depends on capacity constraints, the existence of a three-dimensional loading plan, and battery capacity, which is evaluated under uncertainty through chance constraints. Due to the computational difficulty of compact integrated formulations, we propose a decomposition framework in which a master problem generates candidate routes, while subproblems verify energy and loading feasibility. Two formulations are proposed for the master problem, together with two cut-generation strategies based on logic-based Benders decomposition and the solver's solution pool. The computational validation was conducted using benchmark instances from the literature for road networks with stochastic speeds, extended to account for demands composed of three-dimensional boxes.