A heuristic comparison framework for solving the Two-Echelon Vehicle Routing Problem

The Two-Echelon Vehicle Routing Problem (2E-VRP) is a combinatorial optimization problem used in transport logistics with a two layered composition. In a first level trip, freight has to be delivered from a main depot to a set of possible satellites (intermediate depots). Afterwards the goods are served in second level trips to the final customers. Inspired by a paper from Nguyen, Prins and Prodhon we developed a novel heuristic comparison framework based on the open source simulation environment OpenCI. On the basis of the mentioned paper we implemented three heuristics to solve the 2E-VRP and a greedy randomized adaptive search procedure (GRASP) with a learning process (LP). The GRASP-LP uses two heuristics which improve their solution by a variable neighborhood descent (VND). The VND itself uses two different neighborhoods to enhance the result. The comparison framework is able to load different 2E-VRP maps which can be solved by single heuristics or the GRASP. The new heuristic comparison framework offers the possibility to add quickly different heuristics and to do experiments with them.