Two models of the capacitated vehicle routing problem

The aim of the Capacitated Vehicle Routing Problem (CVRP) is to find a set of minimum total cost routes for a fleet of capacitated vehicles based at a single depot, to serve a set of customers. There exist various integer linear programming models of the CVRP. One of the main differences lies in the way to eliminate sub-tours, i.e. cycles that do not go through the depot. In this paper, we describe a well-known flow formulation of CVRP, where sub-tour elimination constraints have a cardinality exponentially growing with the number of customers. Then we present a mixed linear programming formulation with polynomial cardinality of sub-tour elimination constraints. Both of the models were implemented and compared on several benchmarks.